Quaternary palaeohydrological evolution of a playa lake:Salada Mediana, central Ebro Basin, Spain

BLAS L. VALERO-GARCEÂ S*, ANTONIO DELGADO-HUERTAS  , ANA NAVASà ,JAVIER MACHIÂNà , PENEÂ LOPE GONZAÂ LEZ-SAMPEÂ RIZ* and KERRY KELTS§*Instituto Pirenaico de EcologõÂa CSIC, Apdo 202, 50080 Zaragoza, Spain(E-mail: blas@ipe.csic.es; pgonzal@posta.unizar.es) Estacion Experimental del Zaidin CSIC, Prof. Albareda 18008 Granada, SpainàEstacioÂn Experimental de Aula Dei, CSIC, Apdo 202, 50080 Zaragoza, Spain§Limnological Research Center, U of MN, 220 Pillsbury Hall, Minneapolis MN-55455, USA

ABSTRACT

Sedimentary features, mineralogy, bulk geochemical composition, stable

isotope analyses and pollen data from sediment cores were used to

reconstruct the Late Quaternary depositional evolution of the Salada

Mediana playa lake (central Ebro Basin, northeastern Spain). The 150-cm-

long sediment core sequence is composed of gypsum- and dolomite-rich muds

(Lower and Middle sections) and black, laminated, calcite-bearing sediments

(Upper section). The Salada Mediana formed as a karstic depression in the

Miocene gypsum substratum during the Late Pleistocene. The Lower section

was deposited in a sulphate±carbonate saline lake that ended with a period of

desiccation and basin ¯oor de¯ation. Subsequent deposition (Middle section)

took place in a playa-lake system. Two cycles of lower water table and

expanded saline mud ¯ats occurred. The Holocene sequence is missing,

probably as a result of aeolian erosion. Sedimentation resumed only a few

centuries ago, and saline pan environments dominated until modern times.

The Salada Mediana facies succession was mainly governed by ¯uctuations in

the hydrological balance, brine composition, and salinity; however, aeolian

processes (detrital input and de¯ation) and recycling of previously precipitated

salts also played a signi®cant role.

Keywords Geochemistry, Iberia, Quaternary, saline lakes, stable isotope.

INTRODUCTION

A semiarid climate and the presence of extensiveendorheic areas have favoured the developmentof numerous small saline lakes in Spain. Themajority of them are grouped in four areas: thecentral Ebro Basin, the northern Castille, LaMancha and the Guadalquivir basin (ComõÂn &Alonso, 1988; Pueyo-Mur & De la PenÄa, 1991).Most of the lakes are shallow and ephemeral, andwere formed by karstic or aeolian processes onTertiary evaporitic units. Their chemistry, strong-ly in¯uenced by the substrate, is dominantlysodium chloride or magnesium sulphate. Duringthe past decades, considerable limnological, sedi-

mentological and geochemical work has beencarried out on some modern Spanish lake basins(see Pueyo-Mur & De la PenÄa, 1991 for morereferences), such as La Mancha (De la PenÄa &Mar®l, 1986; OrdonÄez et al., 1994), the centralEbro Basin (Pueyo-Mur, 1979; Davis, 1994; AuqueÂet al., 1995; GarcõÂa-Vera, 1996; SaÂnchez-Navarroet al., 1998), Gallocanta, the largest saline lake inSpain (ComõÂn et al., 1990), and Salines (Giraltet al., 1999).

Knowledge of the sedimentology and hydro-geology of seasonal playa-lake systems has greatlyimproved in recent years with studies in aridregions of Canada (northern Great Plains andBritish Columbia) and Australia (see references in

Sedimentology (2000) 47, 1135±1156

Ó 2000 International Association of Sedimentologists 1135

Smoot & Lowenstein, 1991 and Renaut & Last,1994). In the Iberian Peninsula, however, moderndepositional environments and processes havebeen described for only a small number of lakes,and even fewer have been cored and theirsedimentary sequences analysed in detail (e.g.Gallocanta, Salines, some of the Ebro Basin). Thereduced thickness of sediment accumulated inthese basins, the presence of numerous sedi-mentary hiati, and the complexity of evaporitedeposition and early diagenetic processes havediscouraged the study of these lacustrine basinsas palaeoenvironmental and palaeoclimaterecords. However, the sediments accumulated ingroundwater-fed, discharge playas that experi-ence large ¯uctuations in water level, chemicalcomposition, and salinity are potentially sensitiveindicators of changes in the hydrologic budget(Rosen, 1994). Integrated studies, including sedi-mentology, mineralogy and geochemistry, areneeded to understand the depositional dynamicsof these playa lakes, and to provide models tointerpret their palaeorecords. Saline lacustrinedeposits have helped to reconstruct Quaternarypalaeoclimates worldwide (Renaut & Last, 1994),but the potential of the Spanish saline lakerecords has only recently been acknowledged(Davis, 1994; Giralt et al., 1999). The sediments ofplaya lakes provide the best, and in some casesonly, record of past environmental conditions inthese semi-arid regions. The purpose of this paperis to discuss the modern hydrology and sedimen-tology, and the Quaternary palaeodepositionalevolution of one of these Spanish playa lakes,Salada Mediana, located in a closed-drainage areain the central Ebro Basin, Iberian Peninsula.

METHODS

A 150-cm-long core was collected with a modi®ed5-cm diameter Livingstone corer in the centre ofthe playa lake in August 1996. The sediment corewas split, photographed, described and sampledevery centimetre for organic matter and carbonatecontent (sample size was about 2 g), and every5 cm for other analyses (sample size was about5 g). In all cases, sample thickness was 1 cm.Although care was taken to sample only the mudmatrix and not to include large evaporite crystals(mostly gypsum), small micro crystals could notbe excluded and some were probably sampledtogether with the mud matrix. Facies were iden-ti®ed based on colour, lithology and sedimentarystructures. Organic matter content was deter-

mined by loss-on-ignition analyses at 450 °C,and carbonate contents were measured with acalcimeter. Whole sediment mineralogy was char-acterized using a Siemens D-500 diffractometer;relative amounts (high, medium and low) ofminerals were determined using peak intensity.The magnesium and calcium content of dolomitesand calcites were calculated from the position ofthe main peaks (Goldsmith et al., 1961). Theabsence of iron in the dolomites was inferredusing the lattice spacing of d004 (Goldsmith &Graf, 1958a; Runnells, 1970; Al-Hashimi &Hemingway, 1974). Five core samples (at 3, 16,50, 100 and 150 cm depth) were selected toevaluate the dolomite superstructure re¯ections;silicon was used as an internal standard. Thedegree of ordering of the dolomite samples wascalculated from the relative peak heights of thed015 and d110 re¯ections (Goldsmith & Graf,1958a,b). Clay minerals were identi®ed on ori-ented samples (<2 lm), dried at room tempera-ture, and treated with ethylene-glycol. Carbonand sulphur elemental analyses were performedwith a Perkin-Elmer 260 Analyser. A JSM-6400scanning microscope with EDAX facilities wasused for SEM observations. Bulk sediment sam-ples (0á5 g) were digested with a heated mixtureof HCl and HNO3 (3:1 ratio), ®ltered, and analy-sed for major (Ca, Mg), minor (Al, Sr, Na, Fe, Mn)and trace element (B and Li) composition with aJY 98 inductively coupled plasma spectrometer.Potassium content was measured with a PerkinElmer/Coleman 51-Ca photometer. Samples ofabout 2 cm thick intervals and 25 g weight wereselected every 10 cm for pollen analyses. Pollenwas extracted in 18 samples by chemical attackand ¯otation on Thoulet solution. Pollen countsranged from 180 to 868 grains per sample.

Oxygen and carbon isotopic compositions weremeasured on bulk-sediment samples accordingto standard procedures (McCrea, 1950) and theisotopic values reported in the conventional deltanotation relative to the PDB standard. Mostsamples were composed of only one dolomitephase (stoichiometric, well-ordered, non-ferroandolomite). For samples with a mixed calcite (lowmagnesium and high magnesium phases) anddolomite mineralogy, a double extraction at 25and 50 °C was performed (Al-Aasm et al., 1990).Acid fractionation factors used were 1á01044 at25 °C for the calcite (Kim & O'Neil, 1997) and1á01065 at 50 °C for the dolomite (Rosenbaum &Sheppard, 1986). This separation techniqueshould be valid for the Salada Mediana sedimentsbecause only one dolomite phase is present. The

1136 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

d13C values of organic matter were measured aftercarbonate removal with HCl 1:1. Isotopic compo-sitions of modern sediments, cyanobacterial matsand halophytic plants were also analysed. Watersamples from the lake, the nearby Ginel river, acistern used for cattle, and rainfall were collectedduring the winter and spring of 1997 and theird18O, dD, and major cation and anion composi-tions were measured. Analytical precision wasbetter than 0á1& for d18O and d13C in carbonates,organic matter and water, and better than 2& fordD in water.

GEOGRAPHICAL AND GEOLOGICALSETTING

The saline lakes in the Ebro Basin

The Ebro Basin is a large depression surrounded bythe Pyrenees to the north, the Iberian Range to thesoutheast, and the Catalan Ranges to the east. It ismostly ®lled with Tertiary continental deposits(Quirantes, 1978; PeÂrez et al., 1989; Salvany et al.,1994). In the Salada Mediana area, gypsum andmarls of the Mediana Gypsum Unit (ZaragozaFormation, Miocene) overlie marls and clays of

the Codo Member (Longares Formation, Miocene;Fig. 1). The Mediana Gypsum Unit was depositedin a saline lake ± saline mud ¯at complex, related todistal areas of large alluvial fans originating in theIberian Range during the Lower Miocene (UpperAgenian±Middle Aragonian; PeÂrez et al., 1989).

The central Ebro Basin is the most northerlyarea of truly semiarid climate in Europe. Theclimate is Mediterranean with a strong contin-ental in¯uence, and is characterized by very hotsummers, cold, dry winters, and low rainfall (300±350 mm year)1). The high insolation and evapo-transpiration (1000±1500 mm year)1), and theprevalence of strong, dry, NW winds also contri-bute to a water de®cit through the year, especiallyduring the summer. Rainfall is irregularly distri-buted, although spring and fall precipitationaccounts for >70% of the total annual rainfall.

Lake depressions in the central Ebro Basincommonly occur in groups, particularly in thecentral plateau of Los Monegros, located northof the Ebro river (Fig. 1), where about 100 havebeen described (16 of them ¯ooded every year;Pueyo-Mur & De la PenÄa, 1991). There are sometopographical depressions near Salada Mediana,but none of them are ever ¯ooded. Most salinepans in the central Ebro basin have brines of

EbroBasin

IBERIANPENINSULA

0 25 km

Evaporite Units

Carbonate Units

Sandstone Units

Claystone Units

Ebro RiverQuaternary deposits

X

Saline Mud Flat

Lacustrine terraces

Aeolian and lacustrine sediments

Flat-bottom valleys

Infilled Depression

Pediment and gypsum slopes

Ebro River Terrace

RoadCattleCistern

SaltExploitationRuins

Seepage areas

Coring site

GEOMORPHOLOGY ANDQUATERNARY DEPOSITS

GEOLOGIC LOCATION

MonegrosM

Ebro R.Z

Ginel R

.

M I

O C

E N

E

0 2 km

N

N

41°30´ N

0°44 W

Huerv

a R.

Limestone Unit JURASSIC

AB

C

Fig. 1. (A) Location of the SaladaMediana in the Central Ebro basin,Iberian Peninsula. (B) The playalake is located south of Zaragoza (Z),close to the village of Mediana (M)and lies on the Miocene Evaporiteunit. (C) Map showing the Quater-nary deposits surrounding the lakeand location of the coring site.

Quaternary palaeohydrological evolution of a playa lake 1137

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

(Cl))±(SO�4 )±(Na+)±(Mg2+) type, which undergostrong seasonal oscillations in concentration be-cause of groundwater input, evaporation andprogressive salt precipitation. Several playa lakeswere used during historical times as a source ofsalts. The genesis of the depressions has beenrelated to dissolution of the Tertiary evaporitesubstrate, preferential water circulation throughfaults, differential erosion and surface de¯ation(Quirantes, 1978; Pueyo-Mur, 1979; Soriano,1990; Pueyo-Mur & De la PenÄa, 1991; Benito et al.,1998; SaÂnchez-Navarro et al., 1998).

The Modern Salada Mediana

Geomorphology and sub-environments

The `Salada Mediana' playa lake (latitude: 41°30¢10¢ N, longitude: 0°44¢ W, 350 m a.s.l.) is a small(main axis about 325 m ´ 500 m; 14 ha surface),seasonal (maximum water depth, Zmax � 50±0 cm) playa lake, located 20 km southeast ofZaragoza, on the Mediana Gypsum Unit (Mio-cene; Fig. 1B and 2A). The lake lies in a depres-sion developed on an extensive pediment. Qua-ternary sediments in the watershed correspond tothe remains of Ebro river terraces and the aeolian±alluvial in®ll of ¯at-bottomed valleys composedof gypsarenites and gypsites. Gypsisols developover the alluvial valleys, Sollonchaks around theSalada Mediana, and Cambisols over the pedi-ment (Navas & MachõÂn, 1997). The watershed ofthe Salada Mediana itself is small and has beenaltered by farming during the last decades.Currently, the Salada Mediana has no surfaceinlets or outlets. However, several small, relict,non-functional creeks enter the depression fromthe north and one of them connects with theGinel river valley to the south (Soriano, 1990).Small rounded areas within these valleys maycorrespond to prior depressions ®lled with sed-iments or to alluvial karstic sinkholes (Fig. 1C).The precise age of this drainage system isunknown. The lake and the drainage systemcould have originated during the Lower andMiddle Pleistocene, a period of increased evap-orite solution, responsible for the formation ofmany depressions in the central Ebro valley(Benito et al., 1998). However, most of the de-pressions in the Mediana area formed afterpediment-terrace complex 3 (Soriano, 1990) datedas Upper Pleistocene by the presence of Elephasmeridionales (van Zuidam, 1980). The develop-ment of some of these depressions could havebeen synchronous with the relict drainage sys-

tem, although the Salada Mediana depression isexcavated within the alluvial in®ll, suggesting ayounger age. A preliminary AMS-based chronol-ogy (Valero-GarceÂs et al., 2000) suggests thatsedimentation in the Salada Mediana started afterthe last glacial maximum and most of the sedi-mentary sequence belongs to the late glacialperiod (14 000±10 000 years BP).

Most of the Salada Mediana basin is a seasonalsaline pan (Fig. 2A). A cliff of aeolian and lacus-trine sediments, up to 4á25 m high, occurs on thesouthern and eastern lake margins. At the SWmargin the cliff is higher and is only composed ofaeolian sediments (Fig. 2B). In the SE margin, twowell-developed lake terraces are present at 170 and50 cm above the lake ¯oor, and a small one at 10 cmabove the recent maximum ¯ood level. The ages ofthese aeolian and lacustrine sediments areunknown. A halophytic plant community of Sal-icornia L. and other genera of the Suaedetumbrevifoliae association colonizes the lake margins.The main organic producers in the lake are algalcommunities including Microcoleus desmazieÁres,Dunaliella salina, and diatoms of the Hantzschiagenus. They form mats up to 4-mm thick thatalmost completely cover the lake ¯oor during wetperiods, are detached and reworked by wave andwind action (Fig. 2C), and become partiallyencrusted with evaporites during dry periods.The algal mats are ephemeral features, which arenot annually added to the sediments. Crustaceans(Artemia salina L.), some protozoans and ¯agel-lates characterize the lake biota.

Hydrology and hydrochemistry

The Salada Mediana waters are of (SO�4 )±(Cl))±(Na+)±(Mg2+) type, with low carbonate andcalcium contents, and high Mg/Ca ratios (Table 1).Brine salinity and composition vary greatly dur-ing the year (Mingarro et al., 1981; Auque et al.,1995). The lake is fed by rainfall, groundwaterand runoff. The Salada Mediana is the dischargezone of shallow, uncon®ned aquifers in detritalsediments of the in®lled valleys, the remains ofPleistocene Ebro terraces, and the pediment(Fig. 1C). Diffuse seepage occurs in the south-eastern and northern margins. The lake dries upalmost every year, although it may remain wetduring rainy periods. As in most lakes in theregion, groundwater ¯ow is restricted by the lowpermeability of the Tertiary aquifers and aqui-tards (Auque et al., 1995; GarcõÂa-Vera, 1996), andits contribution to the saline lake water budgetshas not been quanti®ed.

1138 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

Fig. 2. (A and B) The Salada Medi-ana during the ¯ood period (A; win-ter±spring) and during the dry-outperiod (B; summer±fall). Note thecliff developed in the southern edge,the terraces, and the halophyticvegetation (B). (C) Floating cyano-bacterial mats during late spring.

Quaternary palaeohydrological evolution of a playa lake 1139

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

The ultimate source of the solutes in thegroundwaters and playa lakes in the central Ebrobasin is dissolution of Quaternary and Tertiarycarbonates and evaporites (Pueyo-Mur, 1979;Mingarro et al., 1981; SaÂnchez-Navarro et al.,1998). Signi®cant gypsum and carbonate dissolu-tion causes the high SO4

2) and CO32) contents of

the water. The high Mg2+ content of the waterneeds sources in addition to carbonate dissolutionbecause Miocene limestones have a low Mgcontent (1á6% MgCO3) and dolomite- and magnes-ite-rich facies are minor (Quirantes, 1978; Mataet al., 1988). The Miocene evaporite formations(Zaragoza, Lerin, and Falces) contain dolomiteand magnesite associated with clay and evaporitelayers, and also as several-centimetre-thick car-bonate beds (Mata et al., 1988; Salvany & OrtõÂ,1994). The Zaragoza Formation provides Mg to thegroundwaters through dissolution of smallamounts of soluble salts, cation-exchange pro-cesses in the clay fraction, and dissolution ofmagnesite and dolomite. Another signi®cantsource of Mg is the weathering of sepiolite±paligorskite type clay minerals. Mg-rich clayminerals are abundant in the Tertiary Ebro Basin(Torres-Ruiz et al., 1994) and are also present insome playa lakes of Los Monegros (Pueyo-Mur &IngleÂs-Urpinell, 1987a, b). Quaternary sediments(in®lled valleys) and soils also provide a signi®-cant amount of the salts. Indeed, Quaternarysediments in the watershed and the soils containhigh concentrations of Mg2+ (up to 60 meq L)1)and Na+ (up to 20 meq L)1), and signi®cantamounts of K+ (up to 7 meq L)1), and Li+ (up to0á14 meq L)1; Navas & MachõÂn, 1997).

Groundwater and rainfall isotope data for theregion are scarce, and restricted to the centralplateau of Los Monegros. Stable isotope composi-tions (d18O and dD) of groundwaters in the

Monegros endorheic area show high variability(GarcõÂa-Vera, 1996). On average, the deuteriumexcess is very small (�1) and d18O ranges from)7á3 to )4á6& SMOW. The d18O values ofgroundwaters do not increase with TDS, whichsuggests that dissolution is more important thanevaporation for the increase of salinity in ground-waters. The average d18O composition of rain inthe Central Ebro basin is about )7 to )8& SMOW,although available data show a clear seasonality,as expected in semi-arid zones, and a largevariability even during the same month (GarcõÂa-Vera, 1996; Fig. 3). Spring and summer rains haveheavier d18O and dD-values, and a smaller deute-rium excess than fall and winter rains. Thisseasonal variability is controlled by changes inthe isotopic composition of the air masses, andthe kinetic effects associated with evaporation ofthe rainwater during precipitation.

Modern deposition

The geochemical evolution of the Salada Medianabrine shows seasonal cycles governed by tempera-ture and brine concentration (Pueyo-Mur, 1979;Auque et al., 1995). During late autumn, winterand early spring, when the water levels are higherand temperatures colder, the Salada Medianaprecipitation sequence is dominated by mirabi-lite, because of its low solubility at low tempera-tures, although small amounts of carbonate andgypsum also precipitate (Pueyo-Mur, 1979; AuqueÂet al., 1995). Mirabilite dehydrates to thernardite,which forms pseudomorphs after mirabilite andmicrocrystalline aggregates. During the late springand summer, when temperatures are higher, andthe water level decreases, the precipitationsequence changes to carbonate±gypsum±halite.Small amounts of dolomite, calcite, and high-

Ginel spring Ginel river Cistern Salada Mediana

EC 1á68±1á69 3á9±4á2 1á64±1á96 28á3±78á7Cl) 4á9±5á0 10á5±11á6 0á36 120á5±262á4SO4

2) 7á6±9 27á5±29á5 24 360±1500HCO3

) 1á96±2á44 1á33±4á17 1á1±1á71 0á682±2á05Na+ 5á12±5á25 11á4±12á2 0á07±1á06 302á1±1004á8K+ 0á08±0.1 0á15 0á22±0.31 0á7±7Mg2+ 4á82±5á40 5á2±11á8 0á76±1á06 124á8±1015á2Ca2+ 6á78±10á88 29á8±38á0 37á42±47á54 29á6±48á6Sr2+ 0á04 0á2 0±0.06 0B3+ 0á03 0á072±0.087 0á027 0á3±1á12Li+ 0á007±0.009 0á02 0á001±0.0005 0á2±1á1d18O ()7á97)±()8á41) ()7á6)±()7á9) ()4á3)±()9á4) ()6á36)±(+6á75)dD ()18á7)±()60á2) ()57á1)±()61á1) ()43á0)±()61á4) ()56á1)±(+22á3)

Table 1. Electric conductivity(dSm)1), major ion concentrations(meq l)1) and stable isotope (d18O,d2H, SMOW) range of compositionsfrom different water sources in theSalada Mediana area during thesurvey period (February±June 1997).

1140 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

magnesium calcite (HMC) also precipitate, but thebulk of the evaporites is gypsum. Gypsum precipi-tates from the brine as prismatic crystals, 100 lmto several millimetres long. As the water leveldecreases and brine concentrates, halite crystalsalso form in the central area of the salada. Directprecipitation of thenardite and bloedite during thesummer has been documented in the SaladaMediana and other saline lakes in the CentralEbro Basin (Pueyo-Mur, 1979). The mineralogicaldistribution follows a concentric pattern: gypsumand small amounts of thenardite in the marginalareas; towards the inner areas, thenardite dom-inates, although gypsum, bloedite and glauberiteare also present; the central zone consists ofbloedite and thenardite, with small amounts ofhalite (Pueyo-Mur, 1979; Mingarro et al., 1981;Auque et al., 1995). Ef¯orescent salt crusts form inthe saline mud ¯at when the Salada Mediana isdry from early summer until late autumn.

RESULTS

Hydrology and modern carbonate deposition

The water survey (Fig. 3A) shows a rapid chem-ical and isotopic enrichment of lake waters afterthe February rains, and dilution after the mid-April and mid-May rains. Sulphate concentration

increased threefold during the rainless February±April period, whereas Mg2+ increased sixfold,and Cl), Ca2+ and the electric conductivity (EC)increased twofold. The boron and lithium con-centrations in February were already much higherin lake waters than in the river, spring and cistern(Table 1). The cistern, a concrete tank, 6 mdiameter with high walls, is only fed by rainfall.The amounts of B and Li in the lake water sharplyincreased as a result of evaporative concentration,and remained high (about 1 meq L)1) for the restof the sampling period. The mid-April rains onlyslightly decreased the EC. The Ca2+ and Mg2+

concentrations decreased to half of their previousvalues (Mg2+ an order of magnitude more thanCa2+), whereas sulphate, chloride and sodiumconcentrations increased. The Mg2+ and Ca2+

behaviour suggests that the amount of carbonate(dolomite and calcite) precipitated is small, andmost of the Mg2+ is being used up in other phases.

The algal mat sampled in late spring containedevaporites (mirabilite, thenardite, bloedite andhalite), low magnesium and high magnesiumcalcite (LMC 10%, HMC 35% Mg), and clays(illite). The dry algal mat sampled in summeralso contained gypsum, and small amounts ofdolomite and quartz. The position of the XRDpeaks indicates a nearly stoichiometric, non-calcian, non-ferroan dolomite (Goldsmith & Graf,

Fig. 3. (A) Evolution of the major ion concentrations, Mg/Ca ratios, and electrical conductivity (EC) of lake watersduring the survey period (February ± June 1997). (B) Isotopic composition (d18O and dD) of different water sources inthe Salada Mediana area. The Monegros rainfall values (GarcõÂa-Vera, 1996) correspond to the period 1992±94(F, February; M, May; Jn, June; Jl, July; A, August); global, GMWL, and local meteoric water lines (LMWL).

Quaternary palaeohydrological evolution of a playa lake 1141

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

1958b; Goldsmith et al., 1961; Runnells, 1970;Al-Hashimi & Hemingway, 1974).

The Ginel River water isotope composition,electric conductivity and pH remained constantduring the survey period (February±June 1997) atthe two sampling sites, near the village ofMediana and at the headwaters (Magdalenaspring; Fig. 3A and B; Table 1). Immediately aftera rainy period in February, dD compositions ofthe Ginel river, the Magdalena spring, the SaladaMediana and the cistern were similar (about)60&; Fig. 3B). The cistern showed the lightestvalues in both d18O and dD.

Lake waters at the beginning of the surveywere also already chemically concentrated(28á3 dS m)1 electric conductivity) comparedwith the freshwaters of the cistern (1á64dS m)1). After a month without rain (measuredApril 3), the lake waters were enriched by almost12& and 60& in d18O and dD respectively(Fig. 3B), and their conductivity increased to70 dS m)1 (Fig. 3A). The d18O composition ofthe cistern waters only increased by 2&, dDincreased only slightly and conductivityremained in the same range (1á96 dS m)1). Rainyperiods in mid-April and late May decreased

lake water d18O values by 4& and dD by 10&.However, the lake water remained highly con-centrated (electric conductivity 68 dS m)1), andsoon after the rain events, its isotopic composi-tion returned to high values. Although rainfallisotopic compositions were highly variable, val-ues between )9 and )7& are the typical compo-sition of the input waters to the lake. The largeevaporation rates in the area, even during thewinter, and the high surface to depth ratio of theplaya lake, produce a rapid isotopic evolutionthat small rain events cannot counterbalance. Asa result, during carbonate precipitation in springand summer, lakewaters can be enriched by8±12& relative to the average rainwater.

The Quaternary sedimentary sequence

Sedimentary facies

Sedimentary structures (mainly lamination), col-our and lithology allow de®nition of three mainsections in the core (Fig. 4): (i) the Lower section(150±107 cm) is composed of massive grey mudswith intercalated thin gypsum layers; (ii) theMiddle section (107±19 cm) is composed ofmassive to banded (up to several cm thick) grey,

POLLEN

25 50 5 10 10 20 30 25 50

(cm)

UNITS

DEPTH

Evaporite crust

Black, laminated (< 1mm)

0 20 40

CARBONATE

(%)60 0 4 8 12 16

ORGANICMATTER

(%)0 2 4 6

TOTALCARBON

0 4 8

SULPHUR

Sulphate-rich, laminated and bandedcalcite and dolomite-bearing mud

Grey, massive

(%) (%)

MINERALOGY

Clay CalciteDolomite Mg-CalciteQuartz

Cor

ylus

Miri

ophy

llum

Rup

pia

Che

nopo

diac

eae

Dark grey, banded (1 cm)

Grey, red

Gypsum-rich, massive tovaguely banded, dolomite-bearing mud

Gypsum crusts

Gypsum laminae

Isolated gypsum crystals and aggregates

Dark green-grey

FACIES

Massive, microcrystalline,pure gypsum layers(Miocene Zaragoza Formation)

0 2 8 14 20

x 104 grains mg-1

UP

PE

RM

I

D

D

L

E

L O

W E

R

Dark grey, vaguely banded

Gray

Green- Gray

Dark Gray

Gray -Red

LightGray -Red

Dark gray

Black

Gray

FACIES

GYPSUM OCCURRENCES

Low High

(%)(%) (%) (%)

Pollenconcentration Selected species

Low High Low High Low High Low High

Fig. 4. Sedimentary facies, sediment composition (carbonate, organic matter, total carbon and sulphur), semi-quantitative carbonate and silicate mineralogy, pollen concentration, and percentages (without Pinus) of selectedpollen taxa: deciduous tree (Corylus), freshwater aquatic plant (Miriophyllum), saline aquatic plant (Ruppia) andhalophytic grasses (Chenopodiaceae) of the Mediana core.

1142 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

greenish and reddish muds, which contain sev-eral intervals with isolated gypsum crystals; and(iii) the Upper section (19±0 cm) is composed ofdark grey, banded (layer thickness around 1 cm)and black, laminated (lamina thickness <1 mm)muds. Coring was stopped by a massive, micro-crystalline, pure gypsum layer that probablycorresponds to the Miocene Zaragoza Formation.

The core muds are ®ne grained (<20 lm), andmainly composed of gypsum, dolomite, illite andorganic matter. SEM and EDAX observations ofthe salt layer show a mesh of algal ®laments,elongated prisms of magnesium sulphates, clus-ters of gypsum crystals, and some halite crystals(Fig. 5A). Within the muds, gypsum occurs asmillimetre-sized isolated crystals, or euhedral

(prismatic, plates) crystals, tens to hundreds ofmicrons long, and grouped in clusters (Fig. 5B).Dolomite crystals are a few microns long, andanhedral to rounded (Fig. 5D); calcite crystals area few microns long, euhedral to subhedral, andisolated (Fig. 5C). The degree of ordering (Gold-smith & Graf, 1958a,b) in the ®ve measureddolomite samples ranges from 0á6 to 0á9, whichindicates it is well ordered. Dolomite and calcitecrystal sizes are similar (about 5 lm), but thedolomite is more rounded than the calcite(Fig. 5C and D).

The lower sediments (150±121 cm depth) con-sist of alternating massive, structureless mudsand intervals with abundant gypsum laminae andisolated crystals. Gypsum laminae (about 5 mm

Fig. 5. SEM photographs of Mediana sediments. (A) Cyanobacterial mat encrusted with salts (top of unit 1): a meshof algal ®laments (a), with elongated magnesium sulphate crystals (s: mirabilite, thenardite?), gypsum (g) and halite(h). (B) Microcrystalline gypsum: prisms and clusters (unit 5). (C) Calcite crystals (c) at the top of unit 2. Note theeuhedral shapes and smaller size than quartz crystals (q). (D) Dolomite crystals (d) with some clay minerals (cl) at thetop of unit 2. Note the equant, more rounded crystal morphologies of dolomite compared to calcite. Scale bars for allphotographs: 10 lm.

Quaternary palaeohydrological evolution of a playa lake 1143

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

thick) are composed of up to 1 mm long euhedralgypsum crystals with low matrix content.The crystals have no preferred orientation andcontain no matrix inclusions. A thicker gypsum-rich interval occurs between 121 and 107 cm. Thegypsum crusts are composed of long (up to 2 cm)gypsum crystals, with no preferred orientation,and aggregates within a mud matrix. Displacivetextures and matrix mud inclusions in the crys-tals suggest that the crusts formed as a result ofintrasediment growth. The muds from the Middlecore section (107±19 cm depth) are vaguely bed-ded to banded (several centimetre-thick layers).Gypsum occurrences and colour are the mainsedimentary criteria that subdivide this section.Gypsum laminae are absent, but displacive gyp-sum euhedra occur as isolated crystals and crystalaggregates at the base (107±92 cm depth) and atthe top (40±30 cm depth). The lower part of thesection, corresponding to a gypsum-rich interval(107±92 cm depth), is dark-green. The sectionbetween the two gypsum-rich intervals is grey±red, and grades to lighter red colours at the top ofthe section (30±19 cm depth). An abrupt colourtransition occurs at the base of the Upper section.Dark grey, weakly banded muds (19±9 cm depth)change upwards to laminated (<1 mm), blackmuds capped by a salt layer encrusting thecyanobacterial mat.

Sediment composition and mineralogy help tode®ne sedimentary facies and units in the core(Fig. 4). Mud intervals from the Lower sectionhave the highest carbonate (dolomite) content ofthe core. The gypsum-rich intervals (particularlyat the top of the section: 121±107 cm depth) havethe lowest carbonate, organic matter and claymineral content, and also small amounts ofthenardite and bloedite. An abrupt increase inorganic matter and carbonate suggest an uncon-formity between the Lower and Middle sections.The Middle section muds have relatively constantcarbonate (about 10%), organic matter (10%) andsulphate (2±4% sulphur) contents. Two clay-richintervals occur at 80±60 cm depth and at the topof the Middle section (lower part of the Uppersection 30±9 cm depth). The Upper section (19±0 cm depth) is characterized by calcite (high andlow magnesium), and higher contents of mirabi-lite, thenardite and bloedite.

Geochemistry

Figure 6 summarizes the bulk chemical composi-tion of the Mediana core sediments. Calciumprimarily re¯ects the changes in gypsum content,

Fig

.6.

Ch

em

ical

com

posi

tion

of

the

Med

ian

acore

sed

imen

ts.

1144 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

and to a lesser extent carbonate (dolomite andcalcite). As expected, the decreasing trend and thepeak values in the Lower section (Fig. 6) parallelthe sulphur content trends (Fig. 4). However,carbonate-rich intervals are not re¯ected in sim-ilar peaks in calcium, suggesting that dolomite is asubordinate source of calcium. The upward de-crease in calcium in the Middle section correlateswith lower gypsum and dolomite contents, andthe low values (80±60 cm depth interval) correlatewith higher clay mineral contents. The relativeincrease in calcium in the Upper section may alsobe related to the presence of calcite. Strontiumoccurs in sulphates and carbonates (particularlydolomite) as a substitute for Ca2+, and conse-quently it is more abundant in the gypsum- anddolomite-rich facies than in the clay-rich facies.

The other elements show similar patterns: (i)moderate and ¯uctuating values in the Lowersection, (ii) a sharp decline in the gypsum-richinterval (121±107 cm depth), (iii) two cycles inthe Middle section, and (iv) a decreasing trend inthe Upper section. Aluminium, iron and mangan-ese trends follow the clay mineral content, whichsuggests that they are indicators of detrital(aeolian) input. Iron and manganese may beadsorbed on clays, or precipitated as colloidsand oxides. Magnesium shows a slightly differ-ent pattern. It increases from <1% in the under-lying Miocene gypsum to as much as 5% incarbonate-rich intervals of the Lower section, aresponse to the higher dolomite content and thepresence of small amounts of bloedite. Althoughonly illite was detected, the positive correlationwith clay mineral content and negative correla-tion with dolomite in the Middle section suggestthat magnesium is mainly adsorbed to clays,incorporated into Mg-bearing clays, or related tosmall amounts of soluble salts within the muddysediments. Potassium, boron, sodium and lithiumhave similar trends in the sequence, which arerelated to clay and salt content. Sodium does notshow this decrease in the Upper section. After asmall decrease at the base, sodium contentincreases, re¯ecting the presence of halite, mira-bilite, and thenardite in the crust that coveredthe lake bottom during the sampling period.

Pollen

Variations in the most important ecological taxacorrespond to sediment changes in the SaladaMediana core (Fig. 4). The maximum develop-ment of Corylus and other deciduous trees, andthe minimum of halophytic grasses (Chenopodi-

acea) indicate a relatively large arboreal cover inthe area, and a deciduous forest around SaladaMediana during deposition of the Lower section.The presence of Myriophyllum sp., a freshwateraquatic plant, indicates periods of very lowsalinity correlating with the carbonate-rich inter-vals. The demise of Corylus and Myriophyllumsp., and the progressive increase in Chenopodi-aceae, indicative of more arid conditions, aresynchronous with the gypsum-rich interval. Verylow percentages of saline aquatic plants (Rup-pia), the highest Chenopodiaceae content, thelow taxa variability, and the low pollen concen-trations all indicate frequent desiccation eventsand grass colonization of the lake ¯oor in thelower part of the Middle section. A change insediment composition at 60 cm depth correlateswith an increase in Corylus, and a decrease ingrass pollen, congruent with more humid condi-tions and increased lake level. The presence ofRuppia points to more permanent saline watersduring deposition of the upper part of the Middlesection. Decreasing pollen concentration, Corylusand Ruppia percentages, and increasing Cheno-podiaceae indicate more frequent desiccationperiods during deposition of the top part of theMiddle section. The absence of Corylus ± a treethat at present does not grow in the central Ebrovalley ± characterizes the Upper section. Theincrease of Ruppia in the upper part, and thehigh percentages of Chenopodiaceae, favour largeseasonal or pluri-annual oscillation in the waterlevel.

Stable isotopes

There is no comprehensive study of isotopecomposition of the Miocene carbonate forma-tions near the Salada Mediana. In the LosMonegros area, the limestone d18O compositionsrange between )9 and 0& PDB, and dolomite-bearing samples between )6 and +4& PDB(Arenas et al., 1997). The isotopic compositionsof dolomites from the Middle and Upper SaladaMediana sections are within this range, but theLower section has heavier compositions(Fig. 7A). The d18O compositions of Miocenelimestones from the Salada Mediana area, andfrom the Ebro terrace gravels near the lake areconsiderably lighter (between )7 and )4á7&)than calcites from the Upper section (about)2&), and calcite precipitated in the bacterialmats (0á44 and 0á77&; Fig. 7B and C). Unlike thesaline lakes of Los Monegros, where calcite is alsoprovided by detrital local sources (Pueyo-Mur &

Quaternary palaeohydrological evolution of a playa lake 1145

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

IngleÂs-Urpinell, 1987a,b), the restricted occur-rence of calcite, and the isotope data favour anauthigenic origin for calcite in Salada Mediana.

The d13C and d18O values of the dolomite havea strong positive correlation (Fig. 7A). Theygenerally decrease from the bottom to the top ofthe section and there is a sharp negative shiftbetween 9 and 12 cm depth. The d13C calcitevalues decrease downcore in the upper 4 cm(from about )1 to )6&) reaching the lowestvalues in the whole core at the transition betweenthe black and grey laminated muds.

The d13C organic matter curve shows twodistinct populations: (i) samples with d13C val-ues between )25 and )22& (Lower section, andthe 80±60, 45±30 and 19±15 cm depth intervals),and (ii) samples with d13C values between )22and )19& (Fig. 7A). The two main organiccarbon sources in Salada Mediana havedistinctive d13C values similar to these twopopulations (Fig. 8): cyanobacterial mats havevalues between )12á8 and )11á2&, terrestrialhalophytic plants have measured values between)20 to )24&.

DISCUSSION

Hydrology and carbonate formationin Salada Mediana

Brine evolution models for the Central Ebro basinsaline systems based on ®eld observations(Pueyo-Mur, 1979; Pueyo-Mur & De la PenÄa,1991), and experimental studies (Auque et al.,1995), suggest that small amounts of calcite anddolomite precipitate in the Salada Mediana dur-ing the spring and summer, before the main phaseof salt precipitation. Calcite precipitation couldalso be associated with peak algal growth andfreshwater input to the lake during the spring.Although calcite and dolomite are found in themodern cyanobacterial mats, the present data arenot conclusive on modern dolomite formation.The water survey shows that the evolution ofionic concentrations re¯ects a complex interplayof evaporite dissolution and precipitation, soluteintroduction, and brine concentration changes byevaporation and rainfall. The de-coupling ofsalinity and d18O values indicates that factorsother than evaporation and rainfall control the

Fig. 7. Stable isotope composition of the sediments. (A) d18O and d13C values of dolomite samples from the core. (B)d18O and d13C values of dolomite and calcite from the upper sediments (0±9 cm depth). (C) Cross plot of the isotopiccomposition of dolomite from the core, the modern calcite in the cyanobacterial mats, and samples of the Miocenelimestones and limestone clasts from the Ebro river terrace.

1146 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

brine concentration. High chemical concentra-tions were most probably maintained during latespring by the dissolution of previously precipi-tated salts. Furthermore, the isotopic compositionof the brine is more variable than the chemicalcomposition of isotopically depleted and lowsalinity in¯ows. Small in¯ows do not greatlydecrease the brine concentration, and evaporitesformed prior to the rains would buffer brineconcentration. The Salada Mediana illustratesthe complex relationship between salinity andd18O composition observed in other saline lakes(Chivas et al., 1993; OrdonÄez et al., 1994; Valero-GarceÂs et al., 1995, 1997).

The carbonate minerals in Salada Medianahave different origins: (i) aeolian input, (ii)dolomitization, (iii) sulphate-reduction and (iv)precipitation from the lake or interstitial waters.

Aeolian input. Aeolian processes are of greatimportance in the central Ebro basin, as shown bythe occurrence of large wind-blown accumula-tions on the lee side of many saline lakes in theMonegros area (Pueyo-Mur & IngleÂs-Urpinell,1987b; Pueyo-Mur, 1979). The Miocene lime-stones are a negligible source of detrital dolomitebecause it is rare in the Central Ebro area (Quiran-tes, 1978). Dolomite is a minor component asso-ciated with Miocene evaporites and carbonates inthe Ebro Basin (Mingarro et al., 1981; Mata et al.,1988; Salvany & OrtõÂ, 1994; Salvany et al., 1994;Mayayo et al., 1996), and it is also present in someMesozoic and Palaeozoic rocks in the IberianRange. The presence of dolomite in only arestricted number of surface playa lake sediments

in the Ebro Basin (Mingarro et al., 1981) suggeststhat dolomite-bearing formations are only localsuppliers of detrital dolomite and not regionalsources for the Ebro valley. The dolomite-bearingformations in the Salada Mediana watershed arethe Miocene Zaragoza Formation, the Quaternarylacustrine and aeolian deposits surrounding theSalada Mediana, and the saline soils (Solonchakand Gypsisols) developed in the Salada Medianafringes. They probably contribute some aeoliandolomite to the lacustrine sediments. Dolomite isabsent in the Calcisols developed on the Ebroterraces and pediment. De¯ation of Salada Medi-ana lacustrine sediments could be a local source ofdolomite for the aeolian sediments.

Dolomitization. In Salada Mediana, the highlyMg-enriched brines could contribute to earlydolomitization of the calcite, as proposed in otherSpanish saline lakes (Pueyo-Mur & De La PenÄa,1991). The increase in dolomite content withdepth (from 5 to 30%) in several playa lakes inthe Monegros area, and its association withmagnesite were interpreted as a signature of adiagenetic origin for dolomite and magnesite(Pueyo-Mur, 1980; Pueyo-Mur & IngleÂs-Urpinell,1987a,b). In some lakes of the Ebro Basin, wheremore than one core was studied, the largestamounts of dolomite were found in the marginalareas (Monegros lakes: Pueyo-Mur & IngleÂs-Urpi-nell, 1987a,b; and Gallocanta: ComõÂn et al., 1990),where less saline groundwater seepage and mix-ing with the hypersaline lake waters favourdolomitization. The absence of calcite in thelower sections of the Mediana core could indicate

(PDB)‰0 50-5-10-15-20-25-30-35-40

DIC

C3 Plants Atmospheric CO2Soils-CO2

Calculated Calcites

C4 Plants

Medianacarbonates

0°C

30°C15°C

Halophytic Plants Cyanobacterial matsGroundwaters

13CδFig. 8. Sources of carbon in theSalada Mediana: C3 and C4 plants(Deines, 1980); halophytic plantsand cyanobacterial mats;groundwaters from Los Monegros(GarcõÂa-Vera, 1996); atmosphericCO2 (pre-industrial values of )6á5&;current values )8&; Friedli et al.,1986), and soil-CO2 (about 4á5&heavier than the vegetal biomass;Cerling, 1984). Isotopic calculations(DIC and calcites) assumed a calcite-bicarbonate enrichment of 1&(independent of the temperature)and used the calcite-CO2 equation ofRomanek et al. (1992) for tempera-tures of 0 °C, 15 °C and 30 °C.

Quaternary palaeohydrological evolution of a playa lake 1147

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

that replacement of calcite by dolomite occurred.However, the large range in isotopic composition(almost 5&) and the presence of rapid shiftscorrelating with sediment changes are not con-sistent with diagenetic overprint.

Sulphate reduction. Another possible source ofdiagenetic carbonate in sulphate-rich environ-ments such as Salada Mediana is via sulphatereduction. Sulphate-reducing bacteria rapidlyoxidize organic matter to H2S and HCO3

), andthis could lead to the precipitation of calcite withlight d13C values (Kelts, 1988; Komor, 1994). Thed13C dolomite values in the upper 9 cm are fairlyconstant, indicating a relatively stable carbonreservoir after the sharp shift between 1 and 9 cmdepth (Fig. 7B). The d13C calcite values decreasedowncore in the upper 4 cm (from about )1 to)6&), reaching the lowest values in the wholecore at the base of the black laminated muds.These isotopic data may indicate that some HCO3

)

used for calcite precipitation was derived fromdecomposition of organic matter in the reducingupper sediments. Pore waters in Salada Medianaprobably have the oxygen isotopic composition ofevaporated lake waters. The carbon isotopiccomposition, however, could be affected bysulphate reduction and, consequently, calciteprecipitated during early diagenesis would havelighter d13C values. The larger d13C range forcalcite compared with dolomite suggests a doubleorigin for the calcite, primary from the lakewaters, and diagenetic from the pore waters.However, similar d13C calcite and d13C dolomitein the interval below the black muds (4±9 cmdepth), and the consistent 3& shift in d18O valuessupport precipitation of both carbonates from thesame lake or interstitial brine. Organic matter andsulphate concentrations in the upper part of thecore suggest that sulphate reduction may takeplace in the Salada Mediana. Organic mattercontent is moderate (about 8±10%) and decreasesslightly with depth in the upper 30 cm. Totalsulphate content also decreases with depth in theupper sediment section, although generally highorganic matter content in the sediments suggestslow sulphate reduction rates. Lyons et al. (1994)show low sulphate reduction rates in the algalmat ± carbonate platform of Free®ght lake(Saskatchewan, Canada), a similar environmentto Salada Mediana. In conclusion, although sed-imentary and isotopic data suggest that sulphatereduction occurs in Salada Mediana, there is noconclusive evidence that this results in theformation of diagenetic calcite.

Precipitation from lake or interstitial waters.Although carbonate alkalinity is lower, othersedimentological and chemical features of thelake are similar to most modern dolomite-bearinglacustrine environments: shallow playas, salinewaters with high Mg/Ca ratios, NaCl type, andhigh sulphate concentrations (Last, 1990). Theonly study of modern carbonate-forming condi-tions in the Ebro Basin saline lakes was carriedout in Gallocanta (ComõÂn et al., 1990). Theyconcluded that low-magnesium calcite and dolo-mite precipitated from the brine. Modern calcitefrom the Salada Mediana has d13C values similarto, and d18O values heavier than, the uppermostsamples of the Upper section (Fig. 7), indicatinga similar carbon reservoir, but different hydro-logical conditions during the sampling period(probably greater evaporation effects). Althoughdolomite was detected by XRD in the summercyanobacterial mats, the amounts were too smallto measure its isotopic composition. Isotope datastrongly support a primary origin for the moderncalcite and for dolomite and calcite of the Uppersection, either directly from lake water or frominterstitial pore water. Considering that watertemperature during the day in the Salada oscil-lates between 25 and 35 °C, and applying theequation of Kim & O'Neil (1997), the measuredd18O values of modern calcitic (LMC and HMC)sediments formed in the bacterial mats duringlate spring (+0á44 and +0á77&) suggest that calciteprecipitated in isotopic equilibrium with lakewaters (d18O between +2 and +6&).

The oxygen isotopic values of calcites from theupper 9 cm ()2á5 to )1á7&) are about 3& lowerthan dolomite (0á8 to )1á3&). It has been deter-mined experimentally at high temperatures thatdolomite formed from a precursor carbonatefractionates approximately +3& (Land, 1985).Considering the equations of Irwin et al. (1977)and Kim & O'Neil (1997) for dolomite and calcite,respectively, these isotopic compositions supporta cogenetic origin for minerals from the sameevolved waters (between )1 and +3& SMOW attemperatures between 20 and 40 °C; Fig. 9).

The well-ordered structure and stoichiometriccomposition of the Salada Mediana dolomite arenot unequivocal signatures for a detrital origin.About one-third of the Quaternary lacustrineoccurrences listed by Last (1990) are stoichiomet-ric, and about half are well ordered. The betterstudied examples include the west Texas playas(Reeves & Parry, 1965), saline lakes from thenorthern Great Plains of the US and Canada (see

1148 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

references in Last, 1990), and saline lakes inAustralia (De Deckker & Last, 1988, 1989; Rosenet al., 1989). In all cases dolomite was interpretedas a primary precipitate or replacement of acarbonate precursor. Some of the primary dolo-mite occurs as anhedral crystals (western Victorialakes, Australia, and Canadian Plains salinelakes) and rounded aggregates (Free®ght lake,Canada; Last, 1993) similar to the Salada Medianacrystals (Fig. 5).

In summary, although only one mineral phasehas been identi®ed by XRD, dolomite in SaladaMediana probably has several origins. Isotopicand sedimentological data indicate that calciteforms in the modern environment and alsoformed during deposition of the Upper sectionof the sequence. Modern environmental condi-tions and conditions during deposition of theMiddle and Lower core sections were also con-ducive to dolomite formation. Although theaeolian input has not been quanti®ed, there areseveral possible sources of detrital dolomite.

Depositional evolution of Salada Mediana

Palaeohydrological indicators

Sedimentary facies and geochemistry. Thefacies succession was mainly governed by ¯uctu-ations in hydrological balance, brine compositionand salinity; however, aeolian processes (detritalinput and de¯ation), and recycling of salts alsoplayed a signi®cant role. The presence of dunesand aeolian accumulation in many lakes, de¯a-tion of microcrystalline thenardite crystals fromthe lake bed (Pueyo-Mur, 1979), and the presenceof hiati con®rmed by absolute chronologies(Davis, 1994; Valero-GarceÂs et al., 2000) indicate

that aeolian processes contribute greatly to theelimination of salts and sediments from thecentral Ebro Basin saline lakes.

Sedimentary structures, sediment composition,mineralogy and geochemistry allow a moredetailed characterization of the sedimentary facies(Fig. 4). The carbonate and gypsum-rich sedimentsin the Lower section are consistent with depositionin a saline lake setting (Hardie et al., 1978; Lowen-stein & Hardie, 1985; Smoot & Lowenstein, 1991).Intervals in the Middle and Upper sections withhigher Ca, Sr, gypsum and dolomite contents areinterpreted as saline pan deposits. Clay-rich facieswith lower Ca and Sr contents, and higher Al, Mg,Fe and Mn values are interpreted as saline mud ¯atdeposits; Na, K, B and Li also concentrate in thesefacies. The clay-rich intervals could result fromef¯orescent crust development and deposition ofsmall amounts of wind-blown material in depres-sions when the water table was below the sedimentsurface (Smoot & Castens-Seidell, 1994). Periods ofexpansion of the mud ¯ats with more frequentdesiccation stages, ¯uctuating Eh conditions, andhigher aeolian input would be more conducive toiron and manganese precipitation as colloids andoxides. B and Li concentrate in alkaline lake watersat high-salinity stages as evaporative minerals,adsorbed onto the ®ner particles, or substituted forother ions in clay minerals (Kyle, 1994).

Oxygen isotopes. The stable isotope analyses ofmonomineralic carbonate minerals in the lakedeposits provide valuable palaeohydrological andpalaeolimnological information. Although somedetrital input could occur, most dolomite inSalada Mediana is probably authigenic. The largerange (almost 5&) and high d18O values (between+0á8 and +5á47&) of the Mediana dolomite

Fig. 9. Temperature-d18O (water)curves for different d18O (carbonate)compositions, according to theequations of Irwin et al. (1977) andKim & O'Neil (1997). Acid fraction-ation factors used are 1á01044 at25 °C for the calcite (Kim & O'Neil,1997) and 1á01065 at 50 °C for thedolomite (Rosenbaum & Sheppard,1986). Note that dolomites (d18O ofabout +1á3& PDB) and calcites (d18Oof about )2& PDB) from the uppersediment (0±9 cm depth) wouldhave formed in equilibrium withwater of around 0& and +2&SMOW over a temperature range of25±40 °C.

Quaternary palaeohydrological evolution of a playa lake 1149

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

samples (Fig. 7) was expected in a hydrologicalsystem dominated by rainfall, groundwater inputand evaporation. The oxygen isotopic composi-tion of lake water is controlled by (1) the isotopiccomposition of the rainfall, its seasonality, tem-perature and quantity, (2) relative humidity, (3)potential evaporation and (4) groundwater in¯ow(Talbot, 1990; Chivas et al., 1993; Valero-GarceÂset al., 1997). If it is assumed that the bulk of thecarbonate precipitates at approximately the sametime during the annual cycle, when the range ofsurface temperatures is about the same each year,variations in d18O values may primarily representpast changes in the isotopic composition of thewater. However, the decoupling between theisotopic and chemical compositions clearlyindicates that the d18O curve cannot be inter-preted as a function of simple evaporation from aclosed system. Isotopically light waters can beconsistent with increased salinity if lake levelswere so low that annual ¯oods reset the system.

In the absence of major changes in watersources, decreasing d18O suggest a more positivehydrological balance, a colder mean air tempera-ture, a warmer lake-water temperature, or acombination thereof (Talbot, 1990; Valero GarceÂset al., 1997). Large negative d18O shifts, such asthe one at the 12±9 cm depth interval, mostprobably represent an increased input of iso-topically lighter waters. This could be related tohigher effective moisture, a change in the season-ality or rain source, or an increase in groundwaterinput. As cooler precipitation is isotopicallylighter, a decrease in the mean annual tempera-ture or a contribution of more winter precipita-tion would lighten the isotopic composition ofthe lake water. The negative d18O shift at thetransition between the Lower and Middle sec-tions correlates with sedimentological, palyno-logical and geochemical changes that suggest anincrease in lake water level.

The small d18O range in the Middle sectioncould be an indication of an isotopic steady state.The presence of high salt concentrations reducesthe evaporation rate and affects the isotope frac-tionation process, because the isotopic thermody-namic fractionation factor becomes a function notonly of temperature, but also of salinity andcomposition (Gat, 1980; Gon®antini, 1986). Craiget al. (1963) demonstrated that a residual evapo-rated water body reaches a steady-state isotopiccomposition that does not explicitly depend onthe evaporation rate during the ®nal drying-upprocess in conditions of high relative humidity.This steady-state value represents the isotopic

composition of a closed pond under the prevailingclimatic conditions in the area. Consequently, theisotopic response to ¯uctuations in evaporationrate, salinity, and other hydrological parameterscould have been buffered during these periods.This scenario, however, does not fully explainwhy the d18O curve does not show some of thelarge hydrological events, such as ¯ooding andcomplete desiccation, that are recorded by theother sedimentary and geochemical indicators.

Carbon isotopes. Changes in the d13C of authi-genic lacustrine carbonate and lacustrine organicmatter re¯ect variations in the dissolved inorganiccarbon (DIC) pool, controlled by input andbiological processes, mainly respiration andphotosynthesis (HakaÈnsson, 1985; Talbot & Kelts,1990). Fluctuations in groundwater input andcomposition, changes in the limnological andbiological parameters of the lake and in the earlydiagenetic processes are important to the isotopiccarbon budget of lakes (Kelts, 1988; Lyons et al.,1994). The dissolved carbon species in the SaladaMediana come from: (i) groundwater and runoffthat incorporated dissolved carbon throughdissolution of carbonates, decomposition andrespiration of plants, and atmospheric CO2; (ii)equilibration of atmospheric CO2 with the lakewaters; and (iii) oxidation of lacustrine and terres-trial organic matter in the sediments (Fig. 8).There are no groundwater d13C data from theMediana area; the available data from the Mone-gros aquifers show light DIC isotopic composi-tions (d13C between )12á7 and )19á7& PDB) andno correlation with salinity (GarcõÂa-Vera, 1996).Increasing groundwater in¯ow could explain thedecreasing d13C trend during periods of decreas-ing d18O values (12±9 cm depth).

Overall, d13C values of carbonates in SaladaMediana are relatively high ()4á3 to )1á1&)compared with those precipitated in freshwatersystems. In arid environments with poor soils andlow vegetation cover, the in¯uence of atmosphericCO2 vs. plant respiration CO2 in the aquifers andsoils increases, and consequently the d13C DIC ingroundwaters and run-off is relatively enrichedcompared with more vegetated areas (Cerling,1991). Furthermore, the dominance of bacterialmats as the main organic producers in Medianaand evaporative effects could also contribute tothese relatively high values. Cyanobacterialphotosynthesis preferentially extracts 12CO2 fromthe micro-environment and, consequently, car-bonate precipitated in such environments isisotopically heavier than other carbonate precipi-

1150 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

tated in the lake (Andrews et al., 1997). Evapor-ation could also play a role, because extreme 13Cenrichment in evaporite brines have been reportedin the Dead Sea (Stiller et al., 1985). Some of thetrends observed in the d13C record could re¯ectthe interplay of lake productivity and evaporation(residence time evolution). Increased photosyn-thesis produces relatively 13C-enriched carbon-ates, due to productivity-driven enrichment inDIC (Aravena et al., 1992; Meyers, 1994). Longerresidence time and greater evaporation also resultin 13C enrichment (Talbot, 1990; Talbot & Kelts,1990). Decreasing organic productivity in thelower part of the Middle section (107±60 cmdepth), parallel to the increasing dominance ofsaline mud ¯ats, could account for the decreasingd13C trend. Slightly higher d13C dolomite valuesrelated to lower organic matter content (Lowersection, 60±45 cm and 30±19 cm depth intervals)suggest a dominance of residence time evolutionfactors in the carbon isotopic budget. Thedecreasing trend in the 45±30 cm depth intervalcorresponds to increasing organic-matter content,low d13C o.m. values, and a slight increase in d18Odolomite. These indicators are consistent withstronger evaporative effects during periods ofexpanded mud ¯ats.

Carbon isotopic ratios of bulk organic matterprovide palaeolimnological information, such asorganic matter sources, lake palaeoproductivity,and changes in vegetation and catchment hydrol-ogy (Aravena et al., 1992; Meyers, 1994). Thegeneral correspondence of heavier d13C o.m.intervals with saline pan facies (107±70 cm,60±50 cm and 12±0 cm depth intervals), and theabsence of parallel trends in d13C dolomite valuessuggest that changes in the particulate organicmatter and not in the DIC are the main controls ofthe isotopic composition of the bulk organicmatter. The sharp positive shifts in d13C o.m.(up to 4&) are not re¯ected in the d13C dolomitecurve, suggesting that the change in the type oforganic matter in the sediments did not greatlyaffect the DIC from which carbonate precipitated.The different behaviour of the d13C dolomite andd13C o.m. curves could be due to limited post-depositional recycling of organic matter in thesaline pan. Most of the organic matter incorpor-ated into the sediments would be particulate,from cyanobacterial mats and terrestrial plantfragments. The predominant reducing conditionsin the Salada Mediana favour organic matterpreservation and, consequently, the d13C o.m.curve re¯ects the relative input of cyanobacteriallacustrine ()12 to )11&) compared with terrest-

rial ()20 to )24&) organic matter. The negatived13C o.m. excursions during the playa lake in theMiddle section occur in clay-rich saline mud ¯atfacies (60±80 cm and 50±30 cm intervals). Theseintervals correspond with periods of dominanthalophytic vegetation in the Salada Mediana(Fig. 4). The slight increase in d13C dolomite at60 cm is consistent with an increase in lakeproductivity during a saline pan period. However,heavy d13C o.m. compositions during the lowercycle of the Middle section, and increasing valuesin the Upper section correspond with decreasingd13C carbonate values. Sedimentary and geo-chemical indicators point to deposition in areceding saline pan during the top of the Middlesection (30±19 cm depth), although the d13C o.m.are heavier than other intervals interpreted asdeposition in a saline mud ¯at. An increase inlacustrine bacterial and algal organic matter inputduring the saline lake periods was probably themain driving force for the positive d13C o.m.excursions.

The clear covariance between d13C and d18Odolomite values is a re¯ection of hydrologicallyclosed conditions, according to Talbot (1990) andLi & Ku (1997). However, some intervals are morestrongly covariant than others. The Lower sectionof the saline lake deposits, which has the heaviestisotopic compositions, shows the clearest covari-ant trend. Evaporation, increased residence time,and enhanced exchange with atmospheric CO2 inthe saline lake would isotopically enrich watersand DIC. Periods of abrupt and rapid hydrologicalchange in the Salada Mediana, such as the lowerpart of the Upper section (12±9 cm), also showclear covariant patterns, with carbonates depletedin both heavy isotopes. Covariance is weaker inthe playa lake deposits, because of the larger d13Cvariability and the small d18O range.

Depositional and hydrological evolution

Sedimentary facies, elemental composition, sta-ble isotope values and pollen data allow thedetermination of a depositional history for theSalada Mediana. Several AMS dates con®rm aLateglacial age for the Lower and Middle sectionsof the core, and the presence of unsupported210Pb in the Upper section indicates a modern agefor the upper sediments (Valero-GarceÂs et al.,2000). The Holocene sequence has been eroded,probably during some of the mid- and lateHolocene arid periods that occurred in the IberianPeninsula (Valero-GarceÂs et al., 1998). As shownin other lakes in the Ebro basin, sedimentation

Quaternary palaeohydrological evolution of a playa lake 1151

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

resumed only a few centuries ago with the onsetof more humid periods (Davis, 1994; Burjachset al., 1996). The depositional history is dividedinto three cycles, based on the presence ofsedimentary unconformities, and each cycle issubdivided into several subunits based uponpalaeohydrological and palaeoenvironmentalinformation (Fig. 10).

Upper Pleistocene saline lake (Lower section).A karstic depression in the Miocene gypsumsubstratum led to development of a permanentsaline lake. The carbonate-rich muds withintercalated gypsum laminae of the Lower sec-tion were deposited in a permanent sulphate±carbonate saline lake with more diluted(carbonate-dominated) and more concentrated(gypsum-dominated) brine stages. The intervalswith abundant large gypsum crystals and clusters(particularly in the 121±107 cm depth interval)suggest periods of increased intrasedimentgypsum formation in a mud ¯at.

Geochemical indicators (the lowest values ofboron, and relatively low values of lithium andstrontium) and palynological evidence (abun-

dance of Corylus, and presence of Myriophyllumsp.) indicate that carbonate-rich intervals weredeposited in a brackish-lake system, probablythe freshest of the whole sequence. Increasedsalinity during more arid episodes was conduciveto gypsum precipitation from the brine and asef¯orescent crust and intrasediment crystals. Thelow d13C o.m. values suggest that the carbonbudget was not dominated by cyanobacterialmats. Pollen data support the sedimentologicalinterpretation of a deeper brackish lake withoutcyanobacterial mats. High arboreal pollen con-centrations (Fig. 4), indicate a more vegetatedwatershed than later. The high d18O values, andthe strong isotope covariance point to longresidence time, and isotopically evolved waters.Isotopically enriched carbonates during salinelake stages (Lower section) compared with playa-lake stages (Middle section) could re¯ect longerresidence times and greater volumes of evaporatedwater during periods of higher lake level. Isotop-ically heavy, but relatively low-salinity lakewater, could have also been attained by changesin the moisture sources or seasonality (moresummer precipitation). A period of desiccationresulted in the formation of intrasedimentgypsum crusts, and probably sediment de¯ation.

Upper Pleistocene saline pan ± saline mudenvironments (Middle section). The abruptincrease in organic matter and carbonate contentat the onset of the Middle section suggests that anunconformity formed after deposition of thegypsum-rich interval (121±107 cm). The weaklybanded nature of the sediments in the lower partof the Middle section, and the relatively highorganic matter and low carbonate contents, areconsistent with an ephemeral shallow salinelake ± saline pan where evaporite crusts mayhave formed, but were dissolved by the input ofmore dilute waters (saline pan). The increase inlake level allowed colonization of the saline panby bacterial-algal mats, and the surroundingsaline mud ¯ats by halophytic vegetation. Thedominance of bacterial mats as primary producersin the saline pan increased the d13C o.m. values.Higher quartz contents in the lower part (107±100 cm depth) suggest increased run-off duringmore frequent ¯ooded stages.

Small changes in topography and hydrological¯uctuations controlled the extent of the salinepan and saline mud ¯at. The increase in boron,lithium, sodium and clay contents, the abun-dance of Chenopodiaceae pollen, and a sharpdecrease in d13C o.m. values at this interval re¯ect

Fig. 10. Depositional evolution and palaeohydrologyof the Salada Mediana. Dominant lacustrine suben-vironments (shown in the `depositional evolution'column): 1. saline mud ¯ats; 2. saline pan-ephemeralshallow saline lake; 3. permanent saline lake;4. permanent brackish lake.

1152 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

the progressive dominance of desiccation (salinemud ¯at) compared with ¯ooded (saline pan)stages. However, dolomite isotope composition(d13C and d18O) slightly decreased during depos-ition of the lower cycle. Isotopically lighterwaters and increased salinity, as interpreted fromchemical indicators, can be attained in a numberof scenarios: annual ¯oods that reset the systemand dissolve previously precipitated salts, steady-state isotopic processes, increase in the relativegroundwater ratio, and changes in moisturesource or seasonality.

After the arid period recorded in the 60±50 cmdepth interval, saline pan environments domi-nated again. Increased water levels during depo-sition of the lower part of this cycle produced asharp decrease in sodium, boron and lithiumcontent in the sediments, and shrinkage of themud ¯ats. Higher quartz content, and Ruppiapeaks indicate increased run-off during morefrequent ¯ooded stages, and more permanentsaline water. An increase in Corylus also indicatesmoister climatic conditions. The d18O curveshows similar values during this salinity ¯uctu-ation, suggesting low-salinity waters and anisotopic steady-state for the lake. The intervalsof increased isolated gypsum crystals and claymineral content in the middle and upper parts ofthis section (intervals at about 45±30 cm, and atthe transition to the Upper section) indicate morefrequent desiccation periods and increased salinemud ¯at development. Decreased Ruppia contentand an increase in Chenopodiaceae pollencon®rm this trend.

Late Holocene saline pan ± saline lake (Uppersection). The presence of calcite in the Uppersection indicates a different brine composition(higher alkalinity, lower Mg/Ca ratio) whichcould indicate lower salinity. Preservation of®ne lamination, and an increase in quartz con-tent suggest deposition in an ephemeral salinelake or a saline pan with signi®cant periods of¯ooding. Changes in facies and geochemistryindicate the presence of a sedimentary hiatus,which is con®rmed by chronostratigraphicdata (Valero-GarceÂs et al., 2000). The Holocenesequence is missing and the upper sequenceonly re¯ects sedimentation during the last fewcenturies.

Relatively low d13C o.m. values, and an increasein Chenopodiaceae pollen indicate that salinemud ¯ats covered a signi®cant part of the lakebasin. The changes in vegetation and sedimentgeochemistry predated a large negative shift in the

isotopic composition of the brine. The largedecrease in d13C and d18O values in carbonateand the onset of an increasing trend in d13C o.m.indicate a major hydrological and limnologicalchange in the lake during the 12±9 cm interval.Precipitation of calcite, and lighter d18O and d13C(dolomite) values are consistent with an increasein the water level during deposition of the Uppersection, although higher amounts of sulphates(gypsum, bloedite and thenardite), and relativelyhigh values of chemical salinity indicators(sodium, boron) attest to a concentrated brine.The progressive enrichment in the most solublesalts during residual stages of closed basinevolution, and the dissolution of previously pre-cipitated salts, explain higher chemical concen-trations in isotopically lighter waters. A similarincrease in water levels in the recent was noted inseveral lake records in the Ebro basin (Davis,1994), suggesting a regional climatic trend.

CONCLUSIONS

At present, the Salada Mediana consists ofephemeral saline pan and saline mud ¯at envi-ronments, which alternate during the rainy anddry seasons. The modern brine is of Na+ ± Mg2+ ±SO4

2) ± (Cl)) type, sulphate dominated, with lowcarbonate and calcium contents, and high Mg/Caratios. The modern sediments are composed oforganic matter (dominated by cyanobacterialmats), carbonates (dolomite, Mg-calcite, calcite),evaporites (mostly gypsum and mirabilite; sub-sidiary thenardite, bloedite and halite), and aminor silicate fraction (quartz and clays, trans-ported by the wind). Cyanobacterial mats coverthe lake ¯oor almost completely during wetperiods, and become partially encrusted withevaporites and eroded during dry periods. Sedi-mentological and isotopic data favour primaryprecipitation of dolomite and calcite from thelake or interstitial brines, but calcite may alsoform as a result of sulphate reduction. A watersurvey con®rms that evaporation and rainfall arethe main factors controlling the isotopic compo-sition of the lake water.

Sedimentary facies, geochemical and isotopiccompositions and palynological assemblages pro-vide a coherent depositional history of the SaladaMediana. The Lower and Middle sections repre-sent a Late Pleistocene, pre-Holocene sequence.The Lower section was deposited in a sulphate±carbonate saline lake that ended with a period ofdesiccation, and probably basin ¯oor de¯ation.

Quaternary palaeohydrological evolution of a playa lake 1153

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

Subsequent deposition took place in a playa lakesystem dominated by saline pan and saline mud¯at environments. Two cycles of lowered watertable and saline mud ¯at expansion occurred. TheHolocene sediments were eroded, and sedimen-tation resumed only a few centuries ago. Salinepan environments dominated in the Upper sec-tion, where laminated, calcite-bearing, sulphate-rich sediments were deposited.

The Salada Mediana sedimentary and isotopicsequences provide a depositional model forsmall, hydrologically closed Cl)±SO4

� ±Na+±Mg2+ playa lakes that evolved under ¯uctuating,but generally semiarid climatic conditions. Thedifferent depositional environments represent¯uctuations in the water table in response toclimate change. During periods of more frequent¯ooding, the most soluble of the previouslyprecipitated salts dissolved, causing abruptincreases in salinity and changes in brine chem-istry that were not paralleled by the isotopiccomposition of the waters. The de-coupling ofsalinity reconstructions based on bulk-sedimentgeochemistry and the d18O curve underlines theimportance of evaporite recycling in the hydrol-ogy and hydrochemistry of groundwater-fedplaya lakes.

ACKNOWLEDGEMENTS

This research was partly funded by the NationalScience Foundation, USA, under the EAR 941-8657 programme, and by the EEAD-CSIC projectnumber Z-5-96. We are grateful to Loren Hoppeand Concha Arenas for coring assistance. Weappreciate Juan MarõÂn (EstacioÂn ExperimentalAula Dei-CSIC, Zaragoza) for his assistance withSEM observations. We thank Michael Talbot andJoe Smoot for their very thoughtful and compre-hensive reviews, and editor Peter Mozley for®nal revision of the text, which led to a consid-erable improvement of the manuscript. Thisis Limnological Research Center contributionnumber 527.

REFERENCES

Al-Aasm, I.S., Taylor, B.E. and South, B. (1990) Stable isotope

analysis of multiple carbonate samples using selective acidextraction. Chem Geol., 80, 119±125.

Al-Hashimi, W. and Hemingway, J.E. (1974) Recent dolomi-

tization and the origin of the rusty crusts of Northumber-land: a reply. J. Sed. Petrol., 44, 271±274.

Andrews, J.E., Riding, R. and Dennis, P.F. (1997) The stable

isotope record of environmental and climatic signals in

modern terrestrial microbial carbonates from Europe.

Palaeogeogr. Palaeoclimatol. Palaeoecol., 129, 171±189.Aravena, R., Warner, B.G., MacDonald, G.M. and Hanf, K.I.

(1992) Carbon isotope composition of lake sediments in

relation to lake productivity and radiocarbon dating. Qua-tern. Res., 37, 333±345.

Arenas, C., Casanova, J. and Pardo. G. (1997) Stable-isotope

characterization of the Miocene lacustrine systems of Los

Monegros (Ebro Basin, Spain): palaeogeographic and pal-aeoclimatic implications. Palaeogeogr. Palaeoclimatol.Palaeoecol., 128, 133±155.

AuqueÂ, L.F., ValleÂs, V., Zouggari, H., LoÂpez, P.L. and BourrieÂ,G. (1995) GeoquõÂmica de las lagunas saladas de los Mone-gros (Zaragoza). I DeterminacioÂn experimental de los efectos

del reequilibrio mirabilita-solucioÂn con la temperatura en

un sistema natural. Estud. Geol., 51, 243±257.

Benito, G., PeÂrez-GonzaÂlez, GutieÂrrez, F. and Machado, J.(1998) River response to Quaternary subsidence due to

evaporite solution (GaÂllego River, Ebro Basin, Spain). Geo-morphology, 22, 243±263.

Burjachs, F., RodoÂ, X. and ComõÂn, F.A. (1996) Gallocanta:

ejemplo de secuencia palinoloÂgica en una laguna efõÂmera.

In: Estudios PalinoloÂgicos, XI Simposio de PalinologõÂa(Ed. B. Ruiz Zapata), pp. 25±29. Universidad de AlcalaÂ,Alcala de HenaÂres.

Cerling, T.E. (1984) The stable isotopic composition of modern

soil carbonate and its relationship to climate. Earth Planet.Sci. Lett., 71, 229±240.

Cerling, T.E. (1991) Carbon dioxide in the atmosphere: evi-

dence from Cenozoic and Mesozoic paleosols. Am. J. Sci.,291, 377±400.

Chivas, A.R., De Deckker, P., Cali, J.A., Chapman, A., Kiss,E. and Shelley, M.G. (1993) Coupled stable-isotope

and trace-element measurements of lacustrine carbonates

as paleoclimatic indicators. In: Climate Change inContinental Isotopic Records (Eds P.K. Swart, K.C. Loh-

mann, J. McKenzie and S. Savin), Geophys. Monogr., 78,113±121.

ComõÂn, F.A. and Alonso, M. (1988) Spanish salt lakes: theirchemistry and biota. Hydrobiologia, 158, 237±246.

ComõÂn, F.A., JuliaÁ, R., ComõÂn, M.P. and Plana, F. (1990)

Hydrogeochemistry of Lake Gallocanta (AragoÂn, NE Spain).

Hydrobiologia, 197, 51±66.Craig, H., Gordon, L.I. and Horibe, Y. (1963) Isotope exchange

effects in the evaporation of water: 1. Low temperature

experimental results. J. Geophys. Res., 68, 5079±5087.Davis, B.A.S. (1994) Paleolimnology and Holocene environ-

mental change from endoreic lakes in the Ebro Basin, north-east Spain. Unpublished PhD dissertation, University of

Newcastle-Upon-Tyne, UK.De Deckker, P. and Last, W.M. (1988) A newly discovered

region of modern dolomite deposition in western Victoria,

Australia. Geology, 16, 29±32.

De Deckker, P. and Last, W.M. (1989) Modern dolomite incontinental evaporitic playa lakes in western Victoria,

Australia. Sed. Geol., 64, 223±238.

Deines, P. (1980) The isotopic composition of reduced organiccarbon. In: Handbook of Environmental Isotope Geochem-istry, 1. The Terrestrial Environment (Eds P. Fritz and J.Ch.

Fontes), pp. 329±406. Elsevier, Amsterdam.

De la PenÄa, J.A. and Mar®l, R. (1986) La sedimentacioÂn salinaactual en las lagunas de la Mancha: una sõÂntesis. Cuad.Geol. IbeÂrica, 10, 235±270.

1154 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

Friedli, H., Lotscher, H., Oeschger, H. and Stauffer, B. (1986)

Ice core record of the 13C/12C ratio of atmospheric CO2 in the

past two centuries. Nature, 324, 237±238.

GarcõÂa-Vera, M.A. (1996) HidrogeologõÂa de zonas endorreicasen climas semiaÂridos: aplicacioÂn a Los Monegros (Zaragozay Huesca). Publicaciones del Consejo de ProteccioÂn de la

Naturaleza de AragoÂn, Zaragoza.

Gat, J.R. (1980) Isotope hydrology of very saline lakes. In:Hypersaline Brines and Evaporitic Environments (Ed. A.

Nissenbaum), Dev. Sed., 28, 1±7.

Giralt, S., Burjachs, F., Roca, J.R. and JuliaÁ, R. (1999) LateGlacial to Early Holocene environmental adjustment in the

Mediterranean semi-arid zone of the Salines playa-lake

(Alicante, Spain). J. Paleolimnol., 21, 449±460.

Goldsmith, J.R. and Graf, D.L. (1958a) Structural and com-positional variations in some natural dolomites. J. Geol., 66,678±693.

Goldsmith, J.R. and Graf, D.L. (1958b) Relation between lat-

tice constraints and composition of the Ca-Mg carbonates.Am. Miner. 43, 84±101.

Goldsmith, J.R., Graf, D.L. and Heard, H.C. (1961) Lattice

constants of calcium-magnesium carbonates. Am. Mineral.,46, 453±457.

Gon®antini, R. (1986) Environmental isotopes in lake studies.

In: Handbook of Environmental Isotope Geochemistry Vol. 2The Terrestrial Environment, B (Eds P. Fritz and J.Ch.Fontes), pp. 113±168. Elsevier, Amsterdam.

HakaÈnsson, S. (1985) A review of various factors in¯uencing

the stable carbon isotope ratio of organic lake sediments by

the change from glacial to postglacial environmental con-ditions. Quatern. Sci. Rev., 4, 135±146.

Hardie, L.A., Smoot, J.P. and Eugster, H.P. (1978) Saline lakes

and their deposits: a sedimentological approach. In: Modernand Ancient Lake Sediments (Eds A. Matter and M.E.

Tucker), IAS Spec. Publ., 2, 7±41.

Irwin, H., Curtis, C. and Coleman, M. (1977) Isotopic evidence

for source of diagenetic carbonates formed during burial oforganic-rich sediments. Nature, 269, 209±213.

Kelts, K. (1988) Environments of deposition of lacustrine

petroleum source rocks: an introduction. In: LacustrinePetroleum Source Rocks (Eds A.J. Fleet, K. Kelts andM.R. Talbot), Geol. Soc. Lond. Spec. Publ., 40, 3±26.

Kim, S.T. and O'Neil, J.R. (1997) Equilibrium and nonequi-

librium oxygen isotope effects in synthetic carbonates.

Geochim. Cosmochim. Acta, 61, 3461±3475.Komor, S.C. (1994) Bottom-sediment chemistry in Devils Lake,

northeastern North Dakota. In: Sedimentology and Geo-chemistry of Modern and Ancient Saline Lakes (Ed. R.W.Renaut and W. Last), Soc. Econ. Paleont. Miner. Spec. Publ.,50, 21±32.

Kyle, R.J. (1994) Evaporites, evaporitic processes and mineral

resources. In: Evaporites, Petroleum and Mineral Resources(Ed. J. Melvin), Dev. Sedimentol., 50, 477±533.

Land, L.S. (1985) The origin of massive dolomite. J. Geol.Educ., 33, 112±125.

Last, W.M. (1990) Lacustrine dolomite ± an overview ofmodern, Holocene, and Pleistocene occurrences. Earth-Sci.Rev., 27, 221±263.

Last, W.M. (1993) Geolimnology of Free®ght Lake: an unusualhypersaline lake in northern Great Plains of western Canada.

Sedimentology, 40, 431±448.

Li, H.C. and Ku, T.L. (1997) d13C -d18O covariances as paleo-

hydrological indicator for closed-basin lakes. Palaeogeogr.Palaeoclimatol. Palaeoecol., 133, 69±80.

Lowenstein, T.K. and Hardie, L.A. (1985) Criteria for the

recognition of salt-pan evaporites. Sedimentology, 32, 627±

644.

Lyons, W.B., Hines, M.E., Last, W.M. and Lent, R.M. (1994)Sulfate reduction rates in microbial mat sediments of

differing chemistries: implications for organic carbon pres-

ervation in saline lakes. In: Sedimentology and Geochem-istry of Modern and Ancient Saline Lakes (Eds R.W. Renautand W. Last), Soc. Econ. Paleont. Miner. Spec. Publ., 50,13±20.

Mata, M.P., Perez, A. and Lopez-Aguayo, F. (1988) Mineral-ogõÂa del per®l de `La Muela', Terciario del sector central de

la depresion del Ebro (provincia de Zaragoza). Estud. Geol.,44, 135±143.

Mayayo, M.J., Bauluz, B., LoÂpez-Galindo, A. and GonzaÂlez-LoÂpez, J.M. (1996) Mineralogy and geochemistry of the

carbonates in the Calatayud Basin (Zaragoza, Spain). Chem.Geol., 130, 123±136.

McCrea, J.M. (1950) On the isotopic chemistry of carbonatesand a paleotemperature scale. J. Chem. Phys., 18, 849±857.

Meyers, P.A. (1994) Preservation of elemental and isotopic

source identi®cation of sedimentary organic matter. ChemGeol., 114, 289±302.

Mingarro, F., OrdonÄez, S., LoÂpez de Azcona, M.C. and GarcõÂadel Cura, M.A. (1981) SedimentoquõÂmica de las lagunas de

Los Monegros y su entorno geoloÂgico. Bol. Geol. Min., 92,171±195.

Navas, A. and MachõÂn, J. (1997) Assessing erosion risks in the

gypsiferous steppe of Litigio (NE Spain). An approach using

GIS. J. Arid Environ., 37, 433±441.OrdonÄez, S., SaÂnchez Moral, S., GarcõÂa del Cura, M.A. and

Rodriguez Badiola, E. (1994) Precipitation of salts from

Mg2) ± (Na+) ± SO42) ± Cl) playa-lake brines: the endorheic

saline ponds of La Mancha, Central Spain. In: Sedimentol-ogy and Geochemistry of Modern and Ancient Saline Lakes(Eds R.W. Renaut and W. Last), Soc. Econ. Paleont. Miner.Spec. Publ., 50, 61±71.

PeÂrez, A., MunÄoz, A., Pardo, G. and Villena, J. (1989) Evolu-

cioÂn de los sistemas lacustres del margen ibeÂrico de la

DepresioÂn del Ebro (sectores central y occidental) durante el

Mioceno. Acta Geol. Hisp., 24, 243±257.Pueyo-Mur, J.J. (1979) La precipitacioÂn evaporõÂtica actual en

las lagunas saladas del aÂrea Bujaraloz, SaÂstago, Caspe,

AlcanÄ iz y Calanda (provincias de Zaragoza y Teruel). Rev.Inst. Inv. Geol. Diput. Prov. Barcelona, 33, 5±56.

Pueyo-Mur, J.J. (1980) Procesos diageneÂticos observados en las

lagunas de tipo playa en la zona Bujaraloz ± AlcanÄ iz (pro-

vincias de Zaragoza y Teruel). Rev. Inst. Inv. Geol. Diput.Prov. Barcelona, 34, 195±207.

Pueyo-Mur, J.J. and De la PenÄa, J.A. (1991) Los lagos salinos

espanÄoles. SedimentologõÂa, hidroquõÂmica y diageÂnesis. In:

GeÂnesis de Formaciones EvaporõÁticas. Modelos Andinos EIbeÂricos (Ed. J.J. Pueyo), pp. 163±192. Universidad de

Barcelona, Barcelona.

Pueyo-Mur, J.J. and IngleÂs-Urpinell, M. (1987a) Magnesite

formation in recent playa lakes, Los Monegros, Spain. In:Diagenesis of Sedimentary Sequences (Ed. J.D. Marshall),

Geol. Soc. Lond. Spec. Publ., 36, 119±122.

Pueyo-Mur, J.J. and IngleÂs-Urpinell, M. (1987b) Substratemineralogy, pore brine compositions and diagenetic pro-

cesses in the playa lakes of Los Monegros and Bajo AragoÂn,

Spain. In: Geochemistry and Mineral Formation in the EarthSurface (Eds R. RodrõÂguez-Clemente and Y. Tardy), pp. 351±372. CSIC-CNRS, Granada.

Quaternary palaeohydrological evolution of a playa lake 1155

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

Quirantes, J. (1978) Estudio SedimentoloÂgico Y Estratigra®coDel Terciario Continental de Los Monegros. Instituto

Fernando el CatoÂlico, Zaragoza.

Reeves, C.C. and Parry, W.T. (1965) Geology of West Texaspluvial lake carbonates. Am. J. Sci., 263, 606±615.

Renaut, R.W. and Last, W.M. (1994) Sedimentology and geo-chemistry of modern and ancient saline lakes. SEPM Spec.Publ., 50, Tulsa.

Romanek, C.S., Grossman, E.L. and Morse, J.W. (1992) Carbon

isotopic fractionation in synthetic aragonite and calcite:

Effects of temperature and precipitation rate. Geochim.Cosmochim. Acta, 5, 419±430.

Rosen, M. (1994) The importance of groundwater in playas: a

review of playa clasi®cations and the sedimentology and

hydrology of playas. Spec. Pap. Geol. Soc. Am., 289, 1±18.Rosen, M.R., Miser, D.E., Starcher, M.A. and Warren, J.K.

(1989) Formation of dolomite in the Coorong region, south

Australia. Geochim. Cosmochim. Acta, 53, 661±669.

Rosenbaum, J. and Sheppard, S.M.F. (1986) An isotopic studyof siderites, dolomites and ankerites at high temperatures.

Geochim. Cosmochim. Acta, 50, 1147±1150.

Runnells, D.D. (1970) Error in X-ray analysis of carbonates dueto solid solution variation in the composition of component

minerals. J. Sed. Petrol., 40, 1158±1166.

Salvany, J.M. and OrtõÂ, F. (1994) Miocene glauberite deposits

of Alcanadre, Ebro Basin, Spain: Sedimentary and dia-genetic processes. In: Sedimentology and Geochemistry ofModern and Ancient Saline Lakes (Eds R. Renaut and

W. Last), SEPM Spec. Publ., 50, 203±215.

Salvany, J.M., MunÄoz, A. and PeÂrez, A. (1994) Nonmarineevaporitic sedimentation and associated diagenetic pro-

cesses of the southwestern margin of the Ebro Basin (Lower

Miocene), Spain. J. Sed. Res., 64, 190±202.SaÂnchez-Navarro, J.A., PeÂrez, A., Coloma, P. and MartõÂnez-

Gil, F.J. (1998) Combined effects of groundwater and eolian

processes in the formation of the northernmost closed saline

depression of Europe, north-east Spain. Hydrol. Process.,12, 813±820.

Smoot, J.P. and Castens-Seidell, B. (1994) Sedimentary fea-

tures produced by ef¯orescent salt crusts, Saline Valley and

Death Valley, California. In: Sedimentology and Geochem-istry of Modern and Ancient Saline Lakes (Eds R. Renaut

and W. Last), SEPM Spec. Publ., 50, 73±90.

Smoot, J.P. and Lowenstein, T. (1991) Depositional environ-

ments of non-marine evaporites. In: Evaporites, Petroleumand Mineral Resources (Ed. J. Melvin), Dev. Sedimentol.,50, 189±348.

Soriano, M.A. (1990) GeomorfologõÂa Del Sector Centromerid-ional de la DepresioÂn Del Ebro. InstitucioÂn Fernando el

CatoÂlico, Zaragoza.

Stiller, M., Rounick, J.S. and Shasha, S. (1985) Extremecarbon ± isotope enrichments in evaporite brines. Nature,

316, 434±435.

Talbot, M.R. (1990) A review of the palaeohydrological inter-

pretation of carbon and oxygen isotopic ratios in primarylacustrine carbonates. Chem. Geol. (Isot. Geosci. Sect.), 80,261±279.

Talbot, M.R. and Kelts, K. (1990) Paleolimnological signaturesfrom carbon and oxygen isotopic ratios in carbonates from

organic carbon ± rich sediments. In: Lacustrine BasinExploration ± Case Studies and Modern Analogs (Ed. B.J.

Katz), Am. Assoc. Petrol. Geol. Mem., 50, 99±112.Torres-Ruiz, J., LoÂpez-Galindo, A., GonzaÂlez-LoÂpez, M. and

Delgado, A. (1994) Geochemistry of Spanish sepiolite-

paligorskite deposits: Genetic considerations based on trace

elements and isotopes. Chem. Geol. (Isot. Geosci. Sect.),112, 221±247.

Valero GarceÂs, B., Kelts, K. and Ito, E. (1995) Oxygen and

carbon isotope trends and sedimentological evolution of ameromictic and saline lacustrine system: the Holocene

Medicine Lake Basin, North American Great Plains, USA.

Palaeogeogr. Palaeoclimatol. Palaeoecol., 117, 253±278.

Valero GarceÂs, B.L., Laird, K.R., Fritz, S., Kelts, K., Ito, E. andGrimm, E.R. (1997) Holocene climate in the northern Great

Plains inferred from sediment stratigraphy, stable isotopes,

carbonate geochemistry, diatoms and pollen at Moon Lake,

North Dakota. Quatern. Res., 48, 359±369.Valero-GarceÂs, B.L., Zeroual, E. and Kelts, K. (1998) Arid

phases in the western Mediterranean region during the Last

Glacial Cycle reconstructed from lacustrine records. In:Paleohydrology and Environmental Change (Eds G. Benito,

V.R. Baker and K.J. Gregory), pp. 67±80. Wiley & Sons,

London.

Valero-GarceÂs, B.L., GonzaÂlez-SampeÂriz, P., Delgado-Huer-tas, A., Navas, A., MachõÂn, J. and Kelts, K. (2000) Lategla-

cial and Late Holocene environmental and vegetational

change in Salada Mediana, Central Ebro Basin, Spain.

Quatern. Int., (in press).van Zuidam, R.A. (1980) Un levantamiento geomorfoloÂgico de

la regioÂn de Zaragoza. Geographicalia, 6, 103±134.

Manuscript received 16 July 1998;revision accepted 17 March 2000

1156 B. L. Valero-GarceÂs et al.

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 1135±1156

All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.