Hydrogeochemistry of arsenic and other inorganic constituents in groundwaters from La Pampa, Argentina

Applied Geochemistry 17 (2002) 259–284 www.elsevier.com/locate/apgeochem

Hydrogeochemistry of arsenic and other inorganic constituents in groundwaters from La Pampa, Argentina P.L. Smedleya,*, H.B. Nicollib, D.M.J. Macdonalda, A.J. Barrosb, J.O. Tullioc a British Geological Survey, Maclean Building, Wallingford, Oxfordshire, OX10 8BB, UK Instituto de Geoquı´mica (INGEOQUI), Av. Mitre 3100, 1663 San Miguel, Provincia de Buenos Aires, Argentina c Direccio´n de Aguas, Olascoaga 540, 6300 Santa Rosa, Provincia de La Pampa, Argentina

b

Received 3 September 2000; accepted 21 May 2001 Editorial handling by R. Fuge

Abstract Groundwaters from Quaternary loess aquifers in northern La Pampa Province of central Argentina have significant quality problems due to high concentrations of potentially harmful elements such as As, F, NO3-N, B, Mo, Se and U and high salinity. The extent of the problems is not well-defined, but is believed to cover large parts of the Argentine ChacoPampean Plain, over an area of perhaps 106 km2. Groundwaters from La Pampa have a very large range of chemical compositions and spatial variability is considerable over distances of a few km. Dissolved As spans over 4 orders of magnitude (< 4–5300 mg l
1) and concentrations of F have a range of 0.03–29 mg l
1, B of 0.5–14 mg l
l, V of 0.02–5.4 mg l
1, NO3–N of < 0.2–140 mg l
1, Mo of 2.7–990 mg l
1 and U of 6.2–250 mg l
1. Of the groundwaters investigated, 95% exceed 10 mg As l
1 (the WHO guideline value) and 73% exceed 50 mg As l
1 (the Argentine national standard). In addition, 83% exceed the WHO guideline value for F (1.5 mg l
1), 99% for B (0.5 mg l
1), 47% for NO3-N (11.3 mg l
1), 39% for Mo (70 mg l
1), 32% for Se (10 mg l
1) and 100% for U (2 mg l
1). Total dissolved solids range between 730 and 11400 mg l
1, the high values resulting mainly from evaporation under ambient semi-arid climatic conditions. The groundwaters are universally oxidising with high dissolved-O2 concentrations. Groundwater pHs are neutral to alkaline (7.0–8.7). Arsenic is present in solution predominantly as As(V). Groundwater As correlates positively with pH, alkalinity (HCO3), F and V. Weaker correlations are also observed with B, Mo, U and Be. Desorption of these elements from metal oxides, especially Fe and Mn oxides under the high-pH conditions is considered an important control on their mobilisation. Mutual competition between these elements for sorption sites on oxide minerals may also have enhanced their mobility. Weathering of primary silicate minerals and accessory minerals such as apatite in the loess and incorporated volcanic ash may also have contributed a proportion of the dissolved As and other trace elements. Concentrations of As and other anions and oxyanions appear to be particularly high in groundwaters close to low-lying depressions which act as localised groundwater-discharge zones. Concentrations up to 7500 mg l
1 were found in saturated-zone porewaters extracted from a cored borehole adjacent to one such depression. Concentrations are also relatively high where groundwater is abstracted from close to the water table, presumably because this zone is a location of more active weathering reactions. The development of groundwaters with high pH and alkalinity results from silicate and carbonate reactions, facilitated by the arid climatic conditions. These factors, together with the young age of the loess sediments and slow groundwater flow have enabled the accumulation of the high concentrations of As and other elements in solution without significant opportunity for flushing of the aquifer to enable their removal. # 2002 NERC. Published by Elsevier Science Ltd. All rights reserved. 1. Introduction Arsenic is a toxic and carcinogenic element. Health problems from As in groundwater have now been * Corresponding author. Tel.: +44-1491-838800; fax: +441491-692345. E-mail address: [email protected] (P.L. Smedley).

recognised in many parts of the world, including Bangladesh, West Bengal, Taiwan, Inner Mongolia, Mexico, Hungary, Argentina and Chile, as well as more localised occurrences related to mining activity and geothermal sources (e.g. Smedley and Kinniburgh, in press). Symptoms of chronic exposure to As in drinking water at concentrations significantly above 50 mg l
1 are numerous and often severe and include skin, cardiovascular,

0883-2927/02/$ - see front matter # 2002 NERC. Published by Elsevier Science Ltd. All rights reserved. PII: S0883-2927(01)00082-8

260

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

renal, haematological and respiratory disorders. Skin problems are the most commonly recorded and easily recognised and include pigmentation changes (melanosis), keratosis and more serious skin cancers (e.g. Tseng et al., 1968; Chen et al., 1985; Das et al., 1995). Some internal cancers have also been linked to As ingestion, in particular, bladder, lung and prostate cancer (e.g. Smith et al., 1992; 1998). The WHO provisional guideline value for As in drinking water is 10 mg l
1 although many countries, including Argentina, still adopt the former WHO guideline value (1983) of 50 mg l
1 as their national standard. Arsenic problems occur in groundwater because of a combination of its high toxicity at relatively low concentrations (in the mg l
1 range) and its mobility in water in the pH ranges of many groundwaters and over a wide range of redox conditions. Mobilisation of As under reducing conditions in aquifers has been widely documented (e.g. Deuel and Swoboda, 1972; Korte and Fernando, 1991). Mobilisation under oxidising conditions is also known to be an important process and is responsible for As contamination problems in many waters affected by oxidation of sulphide minerals (e.g. Schreiber et al., 2000), often resulting in production of acid-mine drainage (e.g. Nordstrom et al., 2000). Under these conditions, aqueous As can reach very high concentrations, although its mobilisation tends to be comparatively localised because of the strong adsorptive capacity of metal oxides, particularly Fe oxides, in soils and sediments. However, under oxidising conditions at high pH (around 8.5–9.5), As is less strongly bound to Fe oxides than at lower pH values (Dzombak and Morel, 1990). Hence, mobilisation is enhanced and if oxidising, high-pH conditions are regionally extensive in a given aquifer, As contamination may be a widespread phenomenon. The Chaco-Pampean aquifers of Argentina are a good example. Arsenic and other oxyanions have been documented as significant water-quality problems in aquifers from Co´rdoba, Santa Fe and Buenos Aires Provinces (Nicolli et al., 1989; Blarasin et al., 2000; Nicolli and Merino, in press), as well as La Pampa. A number of observed health problems in the region have been linked to As in drinking water. The worst of these are also reported in the provinces of Co´rdoba, Santa Fe and Buenos Aires as well as Santiago del Estero, Chaco and La Pampa. Arsenic-related skin disorders in the region were recorded as early as 1917 and were given the name ‘Bell Ville’ disease after the town in Co´rdoba where the prevalence of symptoms was particularly high (Cı´rculo Me´dico del Rosario, 1917). The condition was described for the first time by Ayerza (1918) as ‘Arsenicismo regional ende´mico’. Keratosis, skin-pigmentation disorders and skin cancer were reported from Co´rdoba Province (Argu¨ello et al., 1938). Biagini et al. (1978) also noted the occurrence of these disorders as well as lung

cancer, mainly from cases in Co´rdoba and Santiago del Estero Provinces. Besuschio et al. (1980) described the increased incidence of several types of neoplasms, particularly of the skin, bladder and digestive tract from medical records of hospital patients in central Argentina, including La Pampa. The As concentrations of the drinking water were reported as 100–1200 mg l
1. Besuschio et al. (1980) found that the most common type of cancer recorded was skin cancer (16.4% for all primary skin carcinomas, excluding melanoma). The most common occurrence was of squamous-cell carcinoma. Hopenhayn-Rich et al. (1996) reported an increased incidence of bladder cancer among populations in Co´rdoba Province drinking water with high As concentrations. They reported standardised mortality ratios for bladder cancer in the same range as values given by other workers for populations in Taiwan. Given the relatively high intake of meat protein (Argentina having one of the highest per capita rates of beef consumption in the world), they suggested that this discounted poor nutritional status as a critical factor in the development of carcinogenic effects. They also discounted genetic factors given the widely differing ethnic backgrounds of populations in Argentina and Taiwan. Although As is likely to be the most serious waterquality problem in the Pampean region, it is not the only problem constituent. High F concentrations have given rise to endemic dental fluorosis. High salinity and concentrations of numerous other trace elements above desirable limits also render the groundwaters of marginal to poor quality for drinking purposes. Understanding the nature and scale of the quality problems is needed as a basis for mitigation. An investigation has been carried out of the groundwaters from northern La Pampa Province to characterise the water chemistry with respect to As and other elements and to determine the principal controlling geochemical factors.

2. Regional setting The Province of La Pampa in central Argentina lies within the Chaco-Pampean Plain, a vast area of around 1.2106 km2 (Nicolli et al., 1989) extending from the Paraguay border in the north to the Patagonian Plateau in the south and to the east of the Pampean Hills. The study area lies in the northern part of La Pampa Province and covers an area of around 11070 km (Fig. 1). The largest town in the study area is Eduardo Castex, some 80 km north of the province’s capital, Santa Rosa. La Pampa experiences a semi-arid temperate climate. Annual rainfall is around 650 mm in Eduardo Castex but increases towards the north and north-east (range 500–700 mm; Michelena and Irurtia, 1995). There is a distinct dry season in the winter (May–September).

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

261

Fig. 1. Sketch map of the Eduardo Castex area showing the topography, main towns, Department boundaries (Rancul, Conhelo, Trenel, Maraco), roads and sampling locations. Locations of the cored boreholes, Talleres Norte and Tamagnoni, are also given.

Records indicate that rainfall totals have been increasing appreciably since the 1970s (Direccio´n de Aguas, Santa Rosa, unpublished data). Average temperatures range between 10  C in winter (July) and 24  C in summer (January). Altitude ranges between over 300 m above sea level in the west to around 150 m above sea level in the east, although numerous small-scale topographic variations exist. Terrain is typically undulating with some sand dunes in the west but with a largely flat plain further east. There are no perennial streams in the area but localised small-scale depressions in the eastern plain fill with water during the wet season (October–April) and undergo partial or complete evaporation during the dry season. Some areas have resulting development of salt deposits around depression margins. The region is predominantly rural and the economy dominated by cattle ranching (mainly in the west) and grain production (mainly in the east). The plains are highly fertile, although the arid climate restricts arable cropping. The lack of perennial surface water sources in northern La Pampa makes groundwater a critical source for public water supply, domestic and agricultural use. In urban areas, public-supply sources are treated by reverse osmosis before use, although a few private boreholes also exist. In rural areas, groundwater is typically abstracted and used without prior treatment.

3. Geology The geology of the Chaco-Pampean Plain comprises a thick sequence of sediments of Mesozoic to Quaternary

age. The topmost part of the sequence consists of blanketing aeolian loess deposits which in places reach several hundred metres in thickness. These form the main exploited aquifers of the region. The Pampean loess deposits are predominantly silts and fine sands and are considered to be mainly of Pleistocene age. During this period, a mountain ice field covered a large area south of 28 S, with winds blowing from the SSW (Clapperton, 1983). North of 28 S, the Argentine Cordillera was a cold, arid environment which allowed efficient production of fine silt by periglacial processes. The loess deposits have been substantially reworked during the Holocene by aeolian and fluvial processes. Well-developed dunes, typically a few metres in height, occur in some areas, particularly in the west (e.g. La Maruja; Fig. 2). The deposits also display abundant evidence of postdepositional alteration, including carbonate cementation and soil-forming processes. The alterations suggest that the region has undergone significant climatic variations during the Quaternary. Conditions appear to have ranged between cold, arid glacial intervals and warmer, wetter interglacial episodes (Kro¨hling, 1999; Tonni et al., 1999). The presence of clayey palaeosols indicates periods of non-deposition and these are considered to have formed during warmer climatic episodes (e.g. Za´rate and Fasano, 1989). Modern soils have largely evolved from the loess sediments and have good permeability, low porosity and deep profiles. They are generally organicpoor. Soil deflation from wind erosion has been significant, particularly in cultivated areas (Michelena and Irurtia, 1995). The loess deposits contain abundant secondary carbonate in the form of calcrete layers (locally known as

262

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

Fig. 2. Model of groundwater flow in the Pampean and deeper aquifers of the study area, showing regional flow gradient and localised zones of discharge in low-lying areas. The distribution of pre-Pampean formations is poorly characterised. In the west (Ing. Foster area) sediments of Triassic age immediately underlie the Pampean sediments. Elsewhere in the section, the Pampean is underlain by Tertiary sediments. Note the vertical exaggeration of approximately 100 times.

‘tosca’), which are often in excess of 1 m thick, as well as concretions, veins or scattered cementing dust. The calcretes may have formed over periods of several thousand years and are an indicator of semi-arid conditions (Za´rate and Fasano, 1989). Calcrete layers are particularly abundant in the east, typically at a depth of around 1 m below ground level. This has resulted in the development of a largely flat-lying plain. Calcrete layers in the west also appear to have had some control on the topography. Volcanic activity has had an important impact on the composition of the Pampean loess. Ash falls were frequent throughout the Tertiary and Quaternary and continue to the present day. The last major ash fall in the region (1932) was from Quiza Pu volcano on the Argentine–Chilean border (Za´rate and Fasano, 1989). A layer of white rhyolitic volcanic ash 0.05–2 cm thick can be traced over much of the study area within the upper part of the loess profile, typically just a few centimetres below surface. In a few places, it has been reworked but usually is maintained as a distinct horizon. The ash is widely considered to be derived from the Quiza Pu eruption.

The distribution of pre-Pampean sediments in the region (Fig. 2) is complex and poorly characterised. Stratigraphic information from borehole logs and geophysical surveys indicates that normal faults are abundant (Elorriaga and Tullio, 2000) and these make regional correlations difficult. Sediments of Triassic to Quaternary age (including the Pampean) infill a SSW to NNE-trending fault-controlled valley in the basement. Basement rocks have been proved in boreholes in the west and central-west parts of the study area at a depth of <80 m, in La Maruja at a much greater depth of 708 m, and beneath Eduardo Castex in the east at a depth of around 180 m (Fig. 2). The basement includes Precambrian metamorphic rocks as well as granite, thought to be of Palaeozoic age. Basement structure has also had some control on topography (Elorriaga and Tullio, 2000). Where recorded, sediments of Triassic age appear to be largely continental red-brown sandy and partly conglomeratic sequences with some intercalations of clay. Both depth to the top of the Triassic sequence, and thickness, vary considerably. In the west around Ingeniero Foster, lithological logs record the presence of an

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

unconformity with deposits of Triassic age immediately underlying Pampean deposits. One log from the western part of the study area (Rancul) records likely Triassic sediments at 120 m depth. Another from the northwestern area [65 050 W 35 330 S] records Triassic red sands and conglomerates with some calcrete at depths between 41–220 m. At La Maruja (Figs. 1 and 2), the top of the Triassic sediments occurs at 276 m depth (Direccio´n de Aguas, Santa Rosa, unpublished data). Other identified sediments in the sequence include silts and sands of presumed Cretaceous age and anhydrite-bearing marine sands and silts of Oligocene or Miocene age (Elorriaga and Tullio, 2000). Given the uncertainties in distribution and stratigraphy of the prePampean sedimentary formations in the study area, these have not been distinguished in Fig. 2.

4. Loess mineralogy and texture The Pampean loess is typically brown or light yellow, sometimes with a reddish or grey tinge (Teruggi, 1957). The deposits are relatively homogeneous and on a local scale are well-sorted. Relative to loess deposits in other parts of the world, the Pampean loess is coarse-grained (modal range being coarse silt to very fine sand). The topmost part of the loess sequence in the study area is largely composed of coarser sandy material, typically 1– 6 m thick. Such upward coarsening of the loess deposits has been noted elsewhere in the Chaco-Pampean Plain (Rabassa and De Francesco, 1986). As noted above, secondary CaCO3 is abundant in the sediments as nodules, layers or cements. Secondary Mn oxide is also present as nodules or cements in some horizons. Petrographic examination of the loess sediments indicates that mineralogy is dominated by plagioclase with variable amounts of quartz, alkali feldspar, often severely altered ferromagnesian minerals, pumice fragments, calcite and heavy minerals (Pearce, 1998). The heavy minerals are dominated by ilmenite but variable amounts of zircon, monazite, apatite and magnetite also occur. Ferromagnesian minerals are typically altered to a chloritic clay and Fe oxide, Ti oxide or mixed Fe,Ti oxide. Clay minerals are dominated by smectite with lesser amounts of illite. No discrete As minerals were identified petrographically.

5. Hydrogeology The groundwater level in the study area is typically around 120 m below ground surface in the west and 4 m in the east, with an overall gradient of ca. 0.0005. However, great variations in water levels occur, particularly in the central region, where recharge is increased

263

locally in the vicinity of sand dunes and decreased in afforested areas (Fig. 2). Seasonal groundwater-level fluctuations are of the order of 1 m. Wells and boreholes range in depth from around 150 m in the west to < 20 m in the east, as a result of the topographically-controlled changes in depth to water table (Fig. 2). Although most abstract groundwater from the Pampean loess aquifer, lithological logs for a few deep boreholes record the presence of fine sands at depth which may be parts of an older stratigraphic unit. In a limited area around Ingeniero Foster in the west, groundwater is abstracted from Triassic sediments below the Pampean strata, with boreholes typically penetrating to depths of 120–150 m (Direccio´n de Aguas, unpublished data). The Pampean aquifer is unconfined as a result of the presence of relatively coarse-grained, sandy deposits in the topmost horizons. Hydraulic conductivities are estimated to be within the upper range (around 10
5 m s
1, ca. 10 m day
1; Herna´ndez and Gonza´lez; 1999) for loess deposits quoted by Freeze and Cherry (1979), although values are lower where the deposits are clayey, such as palaeosol horizons, and where there is limited secondary permeability. Groundwater levels appear to have risen since the early 1970s as a result of increases in rainfall (Direccio´n de Aguas, unpublished data). This has also resulted in the appearance of a number of small lakes and ponds in local depressions. Some of the most persistent of these include one temporary lake in the SE of the study area, close to the Tamagnoni borehole (Section 9.2). The area of this lake varies seasonally but at the time of sampling was around 1.5 km2. A smaller water-filled depression at Santa Rita (Fig. 1) had an area of around 0.05 km2. The occurrence of standing water may be due to perched water tables caused by the presence of extensive, often thick deposits (in excess of 1 m) of calcrete at shallow levels. It may also be an indication of locally reduced infiltration capacity of the unsaturated zone. Where this is the case, discharge from the groundwater system will occur. During the dry season, these areas of standing water can undergo significant evaporation and encrustations of salt deposits, including carbonates, have been produced around the margins of some. In the west, the depths to groundwater are such that there is no groundwater discharge from the aquifer. The annual groundwater recharge has been estimated as 30 mm in the Eduardo Castex area and as 60–100 mm beneath the sand dunes in the west of the study area (using the Cl balance method; Direccio´n de Aguas, pers. commun. 1998). It is likely that much of the rainfall is evaporated or transpired and does not infiltrate below the root zone. Local small-scale topographic variations are likely to have a significant control on groundwater recharge (Logan and Rudolph, 1997). This results in shallow groundwater flow systems with flow directions sometimes in opposition to the regional flow.

264

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

As a result of the semi-arid climatic conditions, low recharge totals and lack of significant topographic variation, groundwater flow in the aquifer is considered to be very slow. Using the regional groundwater-level gradient (0.0005) with parameters for a loess aquifer defined by Herna´ndez and Gonza´lez (1999) and Freeze and Cherry (1979), the travel time for groundwater to flow a distance of 100 km (i.e. from the west to the east of the study area) is estimated to be in the order of 10 ka. Two models outline the influence that the local topographical variations may have on groundwater flow. Where the water table intersects or is close to the surface at a topographic depression, this is likely to be a discharge zone for at least part of the year, with evaporation from the soil or free-water surface and/or transpiration. Where these depressions occur along with local topographic highs, shallow flow cells can be set up. Palaeosols constrain vertical flow (O’Brien et al., 1996) and may play a role in establishing these flow cells, as may the higher permeability of the upper zones of the loess aquifer. Depressions may also be zones of recharge where the water table does not rise to the ground surface and where rainfall is sufficiently intense to cause runoff and ponding of water in the depressions. Localised zones of recharge underlying depressions are characterised by relatively fresh groundwater of low salinity (Bannert, 1974). The appropriate scenario for a given area will depend on the position of the water table in relation to the ground surface and may vary seasonally and with longer-term climatic variation. A model of the groundwater flow system (Fig. 2), involves regional flow from the higher regions of the west towards the east. Superimposed on this regional flow are shallow flow cells which occur in zones of higher recharge where pockets of younger groundwater will be found. To the east, the terrain levels out and groundwater recharge is more diffuse. Here, discharge zones driven by evaporative processes, may cause the recycling of recently recharged water but may also cause deeper, regionallyflowing groundwaters to be brought nearer to the surface. Based on the recharge estimates above, a simple water-balance calculation suggests that the relatively low hydraulic conductivity of the loess does not allow the estimated amounts of water being recharged over the area to flow out of the eastern boundary. Hence, it is possible that a significant proportion of the groundwater flow occurs within deeper, more permeable, aquifers below the loess (Fig. 2). The degree of hydraulic connection between the Pampean deposits and the underlying sediments is not clear. Evidence from some borehole logs in the area and from studies in neighbouring 4Buenos Aires Province (Herna´ndez and Gonza´lez, 1999) suggests that the base of the Pampean formation may consist of poorly permeable deposits and hence connection may be limited.

6. Sampling and analysis Groundwater samples were collected from boreholes in an area covering 110 by 70 km. Sites included 17 public water-supply boreholes and 90 private boreholes or hand-dug wells. Private sources were selected largely at random within the constraints of accessibility and sampling density was about 1 per 50 km2. Surface-water samples were also collected by bailing from two sites, a small pond next to Santa Rita borehole (Fig. 1), hereafter named ‘Santa Rita pond’ [35 59.8300 S 64 22.1420 W] and a lake next to Tamagnoni borehole, hereafter named ‘Tamagnoni lake’ [36 0.4650 S 64 16.6160 W]. On-site analysis included water temperature, pH, Eh (Pt electrode), dissolved O2, specific electrical conductance (SEC, 25  C) and alkalinity (quoted as HCO3). At each site, water samples were also collected for laboratory analysis. Samples for major- and trace-element analysis were filtered (0.45 mm) on site into acidwashed polyethylene bottles. Those collected for analysis of cations and SO4 were acidified to 1% v/v HNO3. Samples for As analysis were acidified to pH 4 (HCl) for analysis of As (III) and subsequently to 2% v/v (HCl) for analysis of total As (AsT). Samples for anion analysis were not acidified. Separate samples were collected for analysis of dissolved organic C (DOC). These were filtered into acid-washed glass bottles using Ag-impregnated 0.45 mm filters and glass syringes. Several samples were collected in duplicate. Unfiltered samples were also collected at some sites for isotopic analysis. Aliquots were collected in glass bottles for 18O, 2H and 13C, in acid-washed polyethylene bottles for 34S, in 1-l polythene bottles for 3H and in large polythene containers (ca. 20 l) for 14C analysis. Aliquots for 34S analysis were prepared by precipitation of BaSO4 and for radiocarbon analysis by precipitation of BaCO3. Major cations, SO4 and trace elements were analysed by ICP–AES (ARL3400 optical-emission spectrometer). A large suite of trace elements was also analysed by ICP–MS (VG PQ1). Arsenic analysis was carried out by ICP–AES with hydride generation (using an ARL 341 hydride generator). Selenium was analysed in a few samples by ICP–AES with hydride generation. All analyses were interspersed with and checked against standard reference materials. Accuracy of most major elements analysed by ICP–AES was found to be generally within 4% and of ICP–MS analyses within 10%. Unacidified aliquots were analysed for NO3–N, NO2–N, NH4–N, Br and I by automated colorimetry. Fluoride was determined by ion-selective electrode. Arsenic(III) was analysed using a modified version of the method given by Driehaus and Jekel (1992). Arsenic(III) was reduced in-line to arsine by NaBH4 at pH>3.5 using 0.5 M acetic acid as buffer solution. Analysis of As(III) was carried out as soon as possible

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

after sample collection (two weeks or less). Total As was analysed following pre-reduction of As(V) to As(III) in samples 24 h before analysis using 5% KI, followed by in-line hydride generation using 1% w/v NaBH4 in 0.1% w/v NaOH (Trafford, 1986). Saturated-zone porewaters were extracted from core sediments by high-speed centrifugation under partial vacuum, using sealed nylon centrifuge cups to minimise CO2 degassing during centrifugation. Extracted porewaters were filtered (0.45 mm) and measurements made of pH, HCO3 and SEC. The porewaters were then separated into two aliquots and one acidified to 1% HNO3 (v/v). Analysis of major and trace elements was then carried out as for the groundwater samples described above, except that anions (Br, I, F) were determined simultaneously by ion chromatography. Where sufficient sample was available, porewaters were also analysed for Se by atomic fluorescence spectrometry (AFS) with hydride generation. Analytical charge imbalances were less than 3% in all but one sample (4.8%). Charge imbalances for porewater analyses were usually less than 2% (the worst being 5%). Stable-isotopic analysis was carried out by mass spectrometry. Results for 18O and 2H are reported relative to SMOW, 13C relative to PDB and 34S relative to CDT. Analysis of tritium and 14C was carried out by liquid scintillation counting. Precision of tritium analyses is better than 1 TU and the detection limit 0.5 TU. Newly cored boreholes were drilled at Talleres Norte in Eduardo Castex and Tamagnoni, 10 km south of Eduardo Castex. Both sites are located in the east of the study area (Fig. 1). The Tamagnoni borehole is sited close to a slight topographical depression where surface water has collected and remains throughout the year, although its volume varies seasonally. Piezometric levels were not determined for the area around the depression but it is likely to be a zone of groundwater discharge, resulting in upward groundwater flow. The Talleres Norte site lies within the Eduardo Castex urban area. Here, the topography is flat and there are no features likely to cause concentrated recharge or discharge. The two boreholes were located next to existing private groundwater supply boreholes that had been sampled previously and were selected because they had high groundwater-As concentrations (Table 1). The cored boreholes were drilled to depths of 26.5 m at Talleres Norte and 30.5 m at Tamagnoni. Near-continuous undisturbed core profiles were recovered by percussion drilling using a hollow core barrel lined with a plastic sleeve. Drilling was carried out without drilling fluid, except where layers of hard calcrete occurred. To penetrate these, a chisel tool was used with drilling fluid (distilled water) and no core was recovered at these intervals. Care was taken to avoid contamination of the

265

recovered core material during the drilling procedure. Further details of the drilling method are given by Smedley et al. (2000).

7. Water chemistry 7.1. Major constituents Representative chemical analyses of selected samples are given in Table 1 and summary statistics for major constituents in Table 2. As a result of significant variations in water levels across the study area, groundwater temperatures are very variable and show a good correlation with well depth as a result of the regional geothermal gradient. Temperatures range between 16 and 29  C (Table 2). Results of chemical analysis show large variations in chemical composition and also indicate the high salinity of many of the groundwaters, with TDS values of 730– 11500 mg l
1 and SEC values in the range 770–17500 mS cm
1. Most are of Na–HCO3 or Na-mixed-anion type, (HCO3–SO4–Cl) although the most saline groundwaters are of Na-Cl type (maximum Cl concentration 4580 mg l
1; Table 2). Sulphate concentrations reach up to 3200 mg l
1 and groundwaters with the highest concentrations approach (though do not reach) saturation with respect to gypsum. The distribution of salinity does not show any clear regional trend but shows much spatial variability over scales of a few kilometres. High salinity values are found in groundwaters in localised discharge zones, though not exclusively. Salinity variations show some trend with depth, high values being found in some of the shallowest groundwaters, most likely as a result of evaporation. Concentrations of TDS in excess of 4000 mg l
1 are more prevalent in groundwaters close to the water table (i.e. where the thickness of the saturated water in the well is around 20 m or less). Carbonate minerals are important components of the mineralogy of the Pampean aquifer and groundwater chemistry is apparently strongly controlled by carbonate reaction. Most of the groundwaters are saturated with respect to calcite and dolomite and a few are supersaturated with respect to dolomite. Alkalinity values are very variable but often high, with a range of 195–1440 mg l
1. Groundwater pHs are neutral to alkaline (7.0–8.7), the highest values corresponding to relatively low pCO2 values, which have minima of around 10
3.0 atm. Calcium varies between 1.6–600 mg l
1 but is mostly low at less than 100 mg l
1. Carbonate reaction of the form: CaCO3 ðsÞ þ CO2 þ H2 O ! Ca2þ þ 2HCO
3 is likely to have been an important process in the groundwaters, with generation of high pHs due to

266

Table 1 Representative analyses of groundwaters from northern La Pampa Sample

a b

m m  S  W 

C

mV mg l
1 mS cm
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mgl
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 % % % %

962083 962085 La Arata Invernada No 3

30 19.8 35.9937 64.3666 27-Oct-96 18.3 8.60 318 6.6 2130 1.55 2.35 526 10.6 12.5 33 1270 5.2 26 < 0.01
2.6 27 73 4800 618 3.4 21.6 183 1.25 8.3 0.3 28.4 4 0.07 180 250 150 0.5 120
5.5
29
12.1

20 34.1 131 5.5 125 35.9655 35.6538 35.7040 64.1872 64.3438 65.1001 27-Oct-96 28-Oct-96 28-Oct-96 18.8 21.0 26.4 8.24 7.88 7.98 321 415 360 4.4 5.2 6.5 2990 2070 1150 4.15 8.4 10.2 7.97 12.2 7.93 695 451 233 15.9 9.6 10.9 70.0 70.0 150 380 250 100 1340 799 258 4.3 1.2 2.2 20 11 <3 < 0.01 < 0.01 < 0.01
2.2
2.0
2.6 27 31 22 46 8 <4 1800 180 8 412 52 132 3.3 0.66 1.8 52.8 38.1 61.1 219 239 188 4.4 0.63 0.090 5.5 3.1 0.84 0.4 0.2 < 0.2 9.01 6.43 1.67 9 6 6 0.33 0.30 0.31 430 264 20 160 210 27 560 190 6.5 0.2 1.4 0.8 60 42 6.2
4.8
27 8.2

SEC: specific electrical conductance. nd: not determined.

962090 Ing. Foster

962099 Talleres Norte 48 4.7 35.9050 64.3011 30-Oct-96 20.4 7.83 406 6.7 6760 24.8 53.0 1360 25.6 970 1300 635 2.9 7 < 0.01
2.1 28 11 210 522 7.8 29.5 2020 0.83 7.8 0.3 6.75 40 2.5 390 640 990 1.6 58

970689 Est. Maria Paula 12

970700 970706 Tamagnoni Est. Enrigue Villalba

14 4.8 35.9267 36.0014 64.3295 64.2699 24-Apr-97 26-Apr-97 17.5 18.6 7.45 8.07 326 359 4.9 3.8 7590 2580 252 5.43 161 8.25 1320 692 16.0 14.8 1780 68.6 620 290 518 1390 140 12 35 15 < 0.02 < 0.02
1.9
2.0 27 27 <3 30 43 1400 41 49 3.6 0.90 72.0 50.1 3670 278 0.26 4.2 3.8 5.3 0.2 0.4 2.92 10.1 b nd nd 2.3 0.27 110 229 34 18 68 160 1.6 0.4 57 120
4.7
5.3
25
27
7.3 8.6

970783 Caleufu

970788 Trenel No. 2

970797 La Maruja No. 2

970801 E. Castex No. 9

82 26.9 30 94 51 39.5 18.8 9.4 22.3 34.8064 35.5978 35.6959 35.6733 35.9216 64.4389 64.5924 64.1312 64.9426 64.2968 27-Apr-97 29-Apr-97 30-Apr-97 01-May-97 02-May-97 23.2 20.0 19.2 26.7 20.3 7.38 6.99 7.82 7.49 7.90 266 282 431 386 0.8 2.4 7.7 5.5 6.4 7830 15000 3570 2910 2980 128 599 17.5 30.5 21.0 164 521 31.1 37.0 23.9 1520 2090 739 546 615 36.4 58.6 14.6 19.8 10.7 1780 4580 385 513 428 1100 630 590 270 470 536 226 655 497 493 92 60 33 12 1.4 166 14 8 <5 <5 < 0.02 < 0.02 < 0.02 < 0.02 < 0.02
1.8
1.8
2.1
1.8
2.3 33 31 28 35 29 <3 <4 6 5 <4 27 <4 200 220 145 30 143 <6 <6 17 77 9.6 < 0.2 0.43 1.46 34.9 132 21.7 57.5 30.5 3100 13000 640 770 587 0.21 0.029 0.62 0.60 0.46 4.3 1.9 4.1 2.1 2.5 0.4 < 0.2 0.3 < 0.2 < 0.2 1.33 0.034 5.03 1.96 7.65 nd nd nd nd nd 3.6 12.0 0.98 0.79 0.92 137 40 268 60 108 31 10 15 9.0 18 43 8.3 170 12 63 3.1 2.0 0.4 0.5 <1 42 17 38 20 23
3.9
4.5
5.1
6.4
4.9
23
26
27
35
27
12.8
10.2
6.9
9.1 7.7 7.8 7.1

970808 Caleufu, El Oton˜o

970817 El Peral

970820 Establ. Don Fidel

28 18.5 25 22.7 2.4 5.5 35.6172 35.5502 35.4170 64.6115 64.4946 64.4644 03-May-97 05-May-97 05-May-97 19.8 19.0 18.4 8.27 7.97 8.05 320 315 322 7.2 6.6 3.9 3910 4070 4050 4.23 32.8 7.78 11.4 28.5 15.0 882 916 1000 14.4 20.9 10.0 281 248 409 440 230 390 1360 1440 1360 8.6 140 58 8 80 8
2.2 27 105 4900 11 < 0.2 25.7 354 2.9 14.0 0.7 9.28 nd 0.67 730 15 730 <2 130
5.2
27
5.8


1.9 27 33 2250 136 1.38 259 13300 1.5 9.5 0.6 13.5 nd 0.90 310 38 110 <2 91
4.3
23
5.9


2.0 26 13 520 20 1.07 62 637 1.2 10.5 0.4 7.73 nd 1.1 366 20 89 <2 140
5.3
31
6.9

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

Well depth Water level Latitude Longitude Sample date Temperature pH Eh DO SECa Ca Mg Na K Cl SO4 HCO3 NO3–N NO2–N NH4–N Log pCO2 Si As(III) AsT FeT Mn Ba Sr V B PT F Se Br I Al Mo Pb U d18O d 2H d13C d34S

962082 Santa Rita

267

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284 Table 2 Summary statistical data for field-collected parameters and major constituents in the Pampean groundwaters Determinand

Units

Min

Max

Median

Mean

10th centile

90th centile

n

Well depth Water level Temperature pH Eh Dissolved oxygen SEC Ca Mg Na K Cl SO4 HCO3 NO3–N NO2–N NH4–N Si

m m  C

6.0 2.1 16.1 6.99 131 0.8 773 1.55 2.01 120 3.3 8.5 6.8 195 < 0.2 < 0.003 < 0.01 21

140 129 29.1 8.66 492 9.9 17500 599 521 3100 71 4580 3200 1440 140 0.17 0.14 39

29.1 14.1 19.8 7.85 325 6.1 2610 20.9 21.5 543 11.8 192 290 653 9.1 0.003 <0.02 29

42 28 20.6 7.82 327 5.9 3340 45.1 45.3 667 14.8 458 430 716 19 0.011 0.01 30

12 4.0 18.1 7.39 242 3.9 1310 4.47 5.97 290 5.78 20 72 400 1.6 <0.005 <0.02 26

95 74 25.2 8.25 425 8.0 6170 100 118 1210 26 1260 920 1240 50 0.020 <0.02 34

103 93 108 108 102 106 108 108 108 108 108 108 108 108 108 108 108 108

mV mg l
1 mS cm
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1

consumption of CO2 and development of correspondingly lower pCO2 values. Like Ca, concentrations of Mg are also generally low (range 2–521 mg l
1; median 45 mg l
1; Table 2). Molar Mg/Ca ratios are very variable, ranging between <1 and 4, but average around 1.8. The groundwaters are universally oxidising with dissolved-O2 concentrations between 0.8 and 9.9 mg l
1 (Table 2) and redox potentials (Eh values) up to 492 mV. Under these conditions, observed concentrations of NO3–N are often high (up to 140 mg l
1), while concentrations of NO2–N and NH4–N are low (Tables 1 and 2). Of the groundwaters investigated, 47% exceed the WHO guideline value for NO3-N in drinking water of 11.3 mg l
1. A significant proportion of the NO3 present is thought to be from evaporation. However, since the correlation between NO3–N and Cl for example (not shown) is weak, evaporation is unlikely to be the sole cause of the high NO3–N concentrations. It is possible that pollution, from fertilisers especially, provides an additional source. As with TDS, highest NO3– N concentrations are observed in shallow depths where the thickness of the unsaturated zone is small (Fig. 3). Occurrence of high concentrations in this zone would be expected from both pollution and evaporative concentration. Concentrations of K are often high (range 3.3–71 mg l
1; Table 2). Some of this may be derived from pollution (e.g. fertiliser), although evaporation is likely to be responsible for many of the higher concentrations. Dissolution of K-bearing minerals and ion-exchange reactions may also be involved.

Fig. 3. Variation in NO3–N with water level (metres below ground level).

Concentrations of Si are in the range 21–39 mg l
1 (Table 2). The range reflects both high groundwater temperatures and reaction of silicate minerals in the aquifer. Many of the silicate minerals in the aquifer are fine-grained (e.g. clays) and easily weathered (e.g. feldspar, volcanic glass). The groundwaters are saturated with respect to quartz, chalcedony and cristobalite. Reaction of silicate minerals is also a key process responsible for the generation of high-pH groundwaters. Reaction of sodic feldspar for example, consumes protons and results in a pH rise. Reaction of albite (NaAlSi3O8) to form kaolinite (Al2Si2O5(OH)4) can be described as: 2NaAlSi3 O8 þ 2Hþ þ 9H2 O ! Al2 Si2 O5 ðOHÞ4 þ2Naþ þ 4H4 SiO4

268

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

The reaction also accounts for the origin of high Na concentrations in the groundwaters. A similar reaction may also produce gibbsite (Al(OH)3) which is found to be saturated in most of the groundwaters. Some dissolved Na may also have been derived by ion exchange. The two surface-water samples from Santa Rita pond and Tamagnoni lake, have low concentrations of dissolved solutes (TDS 150 and 190 mg l
1 respectively). They are oxidising and dominated by Ca and HCO3 ions. No pH determination was made on Santa Rita pond but that of the Tamagnoni sample was 8.1. The low salinity of the samples, which were taken during the wet season, indicates that the surface-water bodies contain a large component of recent recharge, much of which may have been runoff collected in the depressions. However, the enriched stable-isotopic values (18O,2H ; Section 8.1) of the Tamagnoni sample suggests that water in the Tamagnoni lake has undergone an appreciable amount of evaporation, either from the body of open standing water, or by addition of discharging groundwater from the subsurface. 7.2. Iron, manganese and aluminium As the groundwaters are oxidising, solubility of Fe and Mn oxides is low and concentrations of dissolved Fe and Mn are therefore mostly low. Many samples have concentrations below detection limits and most have <300 mg l
1 Fe and < 15 mg l
1 Mn. However, concentrations are higher in a few samples and Fe reaches up to 1160 mg l
1 and Mn up to 79 mg l
1 (Table 2). At the pH of the groundwaters, the few high concentrations are most likely to be due to presence of colloidal particles which passed through the 0.45 mm filter. Most samples also have less than around 200 mg l
1 Al, although a few have high concentrations (up to 990 mg l
1; Table 2). In these, Al correlates with Fe. The sample of filtered surface water collected from Santa Rita pond, which was visibly cloudy, has the highest

concentrations of Fe, Mn and Al and reflects the dispersion of colloids (< 0.45 mm) at low salinity. Although a few samples appear to have a colloidal component which can have a major effect on concentrations of trace elements due to adsorption reactions, most have negligible concentrations of colloids. Colloidal transport is therefore not thought to have a major impact on the regional groundwater chemistry. 7.3. Arsenic Arsenic concentrations in the groundwaters vary over 4 orders of magnitude (< 4–5300 mg l
1, median 150 mg l
1). Of the 108 groundwaters collected, 95% exceed the WHO guideline value of 10 mg l
1 and 73% exceed the Argentine national standard of 50 mg l
1. Concentrations show little distinct regional trend (Fig. 4) but as with many other determinands, display a significant short-range spatial variability. Of the 6 samples taken from the western area around Ingeniero Foster and possibly deriving from an older Triassic aquifer rather than the Pampean, only one contains As at a concentration less than the WHO guideline value (i.e. with a value of 8 mg l
1). This sample has a relatively low HCO3 concentration (258 mg l
1) but has otherwise similar characteristics to groundwaters from the Pampean aquifer (Na–HCO3 dominant, high Cl, SO4). It is likely, though not certain, that groundwater from this location is hydraulically connected to that from the overlying Pampean aquifer. Arsenic shows no correlation with Fe, Mn or Al, although the few samples with high concentrations of these elements (taken to be colloidal) have relatively low As concentrations. These comprise only around 5 groundwater samples and the surface-water sample from Santa Rita pond. In the remainder of the samples, colloidal influence is thought to be negligible. As a result of the oxidising condition of the groundwaters, As(V) (arsenate) is the dominant species present.

Fig. 4. Map of northern La Pampa showing distribution of As concentrations.

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

269

Fig. 5. Variation of As(III)/AsT ratio with AsT concentration in the Pampean groundwaters.

Arsenic(III)/AsT ratios are typically low at around 0.017 (Fig. 5) and the maximum observed As(III) concentration is 110 mg
1 (for an AsT concentration of 4900 mg l
1). Dissolved As shows a positive correlation with pH (r2 0.46, 95% confidence limit) and alkalinity (HCO3, r2 0.64; Fig. 6). The high pHs generated by carbonate and silicate reaction are thought to be the dominant control on As mobilisation. At high pH, arsenate sorption onto Fe oxides in particular (but also Mn oxides) is weakest (e.g. Dzombak and Morel, 1990). Correlations are also notable between As and other anions and oxanions, especially F (r2 0.70), V (r2 0.85), B, Mo, U and to a lesser extent Be (Fig. 6). Similar correlations between As and F and V in particular have also been observed in groundwaters from neighbouring Co´rdoba, Santa Fe and Buenos Aires Provinces (Nicolli et al., 1989; Nicolli and Merino, in press; Blarasin et al., 2000). Although the distribution of As in the groundwaters is highly variable, there is a tendency for some of the high concentrations to occur in low-lying areas (localised depressions). Examples include groundwaters from the low terrain around Tamagnoni borehole, as well as Santa Rita (Fig. 1). The reasons are not entirely clear, but are considered to be related to the local patterns of groundwater circulation. Flow models (Section 5) suggest that the depressions are the centres of convergent localised flow and act as discharge zones during periods of high groundwater level. Although potentially diluted by surface recharge during wet periods, the depressions are likely to be zones of slow groundwater movement and lack of exit of groundwater. In this case, the degree of aquifer flushing and associated removal of dissolved ions will be restricted. High pH values in these zones can also be generated by dissolution of calcite in the aquifer with a corresponding fall in pCO2 levels. Highest concentrations of As also tend to be found in wells where the saturated well thickness is small, i.e. those wells which penetrate only a few metres below the

water table (Fig. 7). It appears that more As release has occurred in the zone of water-table fluctuation than at greater depths. This increase at shallow depths is not simply a result of evaporation however, since there is no positive correlation between As and indicators of salinity such as Cl (Fig. 8). The correlation between As and Cl is weak but if anything is negative rather than positive. Hence, factors other than evaporation, such as desorption from metal oxides and possibly silicate reaction, are likely to be controlling As mobilisation. Speciation modelling (PHREEQC using the WATEQ4F database) suggests that the groundwaters are undersaturated with respect to As minerals, although barium arsenate (Ba3(AsO4)2) is calculated to be supersaturated in many. This mineral has not been identified in the loess deposits and it is likely that the values reflect inadequately known thermodynamic data for the mineral (Cullen and Reimer, 1989) rather than true supersaturation. 7.4. Other anions and oxyanions As described above, elements which form anions and oxyanions in solution, notably F, V, B, Mo, U and Be, generally have good positive correlations with As. In solution, these are likely to occur dominantly as fluoride, vanadate, borate, molybdate and uranyl-carbonate species which, like As, are preferentially mobilised at high pH. Beryllium is likely to occur in solution as Be(OH)2(aq) and Be(OH)
3 ions and may form complexes with F (Hem, 1992). Mutual competition between the oxyanion trace elements as well as HCO3 for available adsorption sites is a potential additional factor which may contribute to high dissolved concentrations in the groundwaters with the highest pH values. Of the anion and oxyanion-forming elements, highest concentrations are observed for F which occurs in solution mainly as F
. Speciation modelling suggests that fluoroarsenate complexes are minor, even in samples with the highest observed As and F concentrations.

270

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

Fig. 6. Variation of AsT concentrations with pH, HCO3, F, V, B, Mo, Be and U in the Pampean groundwaters. Plotted trace-element concentrations are uncensored; for AsT the analytical detection limit was 4 mg l
1 (6). Where applicable, WHO guideline values (GVs) are also shown. Linear regression curves are also given with the corresponding correlation coefficients (r2). Confidence limits in each case are 95%.

Concentrations of F range between 0.03 and 29.2 mg l
1 and 83% of groundwaters investigated exceed the WHO guideline value of 1.5 mg l
1 for F in drinking water (Table 3). Many of the high-F groundwaters are saturated with respect to fluorite. Fluoride correlates particularly well with As (Fig. 6). Concentrations of B range between 0.46 and 14 mg l
1. The WHO (1998) revised guideline value for B in drinking water is 0.5 mg l
1. Of the samples investigated,

all but one exceed this value (99% exceedance; Table 3) and many are significantly in excess. The correlation between B and As is not as strong as some other elements given in Fig. 6. Vanadium varies between 0.02 and 5.43 mg l
1 (Table 2). Such concentrations are extremely high by world groundwater standards. Concentrations in groundwaters are typically < 1 mg l
1 (Wanty and Goldhaber, 1992) and usually much less. Vanadium is

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

271

Fig. 7. Variation of Cl, SO4, AsT, V, F and B in the Pampean groundwaters with saturated depth (total well depth minus water level, in metres).

likely to be dominantly present in solution as vanadate
(HVO2
4 , H2VO4 ) ions. Vanadium shows a strong positive correlation with As (Fig. 6) and is likely to derive from similar mineral sources (secondary Fe and Mn oxides as well as magnetite and ilmenite) under the highpH conditions. Molybdenum varies in concentration between 2.7 and 990 mg l
1 (Table 2). Of the sources investigated, 39% exceed the WHO guideline value of 70 mg l
1. Molybdenum is weakly correlated with As (Fig. 6). Uranium concentrations range from 6 to 250 mg l
1. All samples exceed the revised WHO (1998) guideline value for drinking water of 2 mg l
1; (Table 3). At the pHs of the groundwaters, the dominant dissolved species are likely to be uranyl carbonate complexes 4
(UO2(CO3)2
2 , UO2(CO3)3 ). A good positive correlation is observed between U and alkalinity (HCO3) in the groundwaters (as for the other oxyanions).

Fig. 8. Variation of As with Cl concentration in the Pampean groundwaters.

Beryllium concentrations vary between < 0.01 and 0.4 mg l
1. The absolute concentrations are low but are relatively high compared to world-average groundwaters. Beryllium shows a weak positive correlation with As (Fig. 6).

272

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

Table 3 Statistical summary of trace-element and stable-isotopic data for the Pampean groundwaters Units Ag Al As(III) AsT B Ba Be Bi Br Cd Ce Co Cr Cs Cu Dy Er Eu F Fe Ga Gd Ge Ho I La Li Lu Mn Mo Nd Ni Pb Pr P Rb Sb Se Sm Sr Tb Th Tl Tm U V Y Yb Zn

mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1

d18O d2H d13C d34S

% % % %

a

Min <0.04 2.9 <3 <4 0.46 5.0 <0.01 <0.02 0.06 <0.02 <0.008 <0.03 0.4 <0.01 0.4 <0.007 <0.007 <0.004 0.034 <6 <0.01 <0.01 <0.02 <0.002 17.4 <0.013 5.7 <0.002 <1 2.7 <0.009 <0.2 <0.45 <0.004 <0.2 1.5 <0.05 <2 <0.01 0.066 <0.003 <0.004 <0.01 <0.002 6.2 0.02 <0.002 <0.006 2.5
6.4
36
12.8 6.3

Max 0.04 990 110 5300 13.8 260 0.40 <2 12 2.7 1.7 1.3 20 0.22 89 0.098 0.051 0.03 29.2 1160 0.18 0.14 0.45 0.011 730 6.4 150 0.004 79 990 0.49 19 14 0.16 0.7 32 0.92 40 0.11 13.3 0.02 0.45 0.14 0.008 250 5.4 0.51 0.039 1400
3.8
23
4.0 8.7

Median

Meana

16 4.2 150 3.0 36 0.01

55 11 414 3.5 45 0.05

6.9 <4 20 1.5 19 <0.09

110 28 730 6.4 81 0.11

0.67 0.11 < 1.7 0.13 2.3 0.02 7.1

1.2 0.28 0.15 0.21 3.0 0.10 12.0

0.15 <0.1 <0.5 0.02 1.0 <0.4 1.3

2.5 0.46 0.12 0.45 5.6 0.06 28

3.84 55 <2

5.24 125 0.09

1.30 10.0 <0.4

9.94 337 0.01

10th centile

90th centile

n

58

110

19

260

106 107 108 108 108 108 107 106 108 107 107 107 107 107 107 107 107 107 108 108 107 107 107 107 108 107 107 107 108 107 107 107 107 107 108 107 107 34 107 108 107 107 107 107 107 108 107 107 107


4.8
27
8.5 7.9


4.8
27
8.1 7.9


5.3
32
11.6 7.1


4.4
24
5.0 8.6

60 60 34 20

< 2.4

0.14

<0.5

0.17

121 < 0.8 26

159 0.16 32

40.0 <0.3 12

292 0.13 57

2.0 62

5.2 110

<1 9.2

11 180

0.88 0.39 < 0.2 5.0 8

2.0 0.82

<0.3 <1.3

4.4 1.2

0.2 6.4

<0.2 2.6

0.4 12

12

3

0.60

1.3

0.005

0.09

<0.3 <0.5

0.09 0.04

42 0.84 0.28

17 0.16 <0.1

82 1.7 0.13

31 0.56 0.03

0.20

27 3.0

Mean values were calculated by substituting a value of half the detection limit where values were found to be below the detection limit. Median and mean values have not been quoted where a significant number of samples were below detection limit.

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

Despite its preferential mobility as chromate under oxidising conditions, Cr concentrations are usually low in the groundwaters, reaching up to 20 mg l
1. This is likely to be because the loess deposits are derived mainly from Andean igneous rocks (dacite, rhyolite, andesite) which are typically Cr-poor (e.g. Middlemost, 1985). Chromium shows no linear relationship with As. Concentrations of P are high in some samples, the range varying between < 0.2 and 0.7 mg l
1, although the median value is low ( < 0.2 mg l
1; Table 3). There is no apparent relationship of dissolved P with As. Selenium was only measured in a few of the pumped groundwater samples. Concentrations were found to be between <2 and 40 mg l
1, with 32% of samples measured exceeding the WHO guideline value of 10 mg l
1. Such values are relatively high. Concentrations in groundwaters of similar chemical composition from neighbouring Co´rdoba and Santa Fe Provinces were also found to have some high Se concentrations (up to 24 mg l
1 in Co´rdoba; Nicolli et al., 1989 and up to 160 mg l
1 in the Carcaran˜a´ Basin of Co´rdoba and Santa Fe; Nicolli and Merino, in press). The Se is likely to be 2
present as selenate (SeO2
4 ) or selenite (SeO3 ) in solution. Unlike the other oxyanions, dissolved Se does not appear to show any correlation with pH, HCO3 or As. It shows a weak positive correlation with groundwater salinity (e.g. Cl, SO4, SEC). Associations were also found between groundwater Se and salinity by Nicolli et al. (1989) for Co´rdoba Province. Similar associations have been found in saturated soils from the San Joaquin Valley, California (Fujii et al., 1988). 7.5. Other trace elements As with Cr, concentrations of many of the other transition metals are relatively low. Nickel has a median concentration of 0.9 mg l
1 although it reaches up to 19 mg l
1 (close to the WHO guideline value of 20 mg l
1 for drinking water). Cobalt reaches a maximum of 1.3 mg l
1 and Cd of 2.7 mg l
1, with most samples being less than 0.5 mg l
1. The low concentrations of these elements are also believed to reflect the acidic (silica-rich) composition of the loess sediments and ashes. As these elements form cationic species in solution, their mobilisation is not favoured under the alkaline conditions of the groundwaters. Concentrations of Cu and Zn reach up to 89 and 1400 mg l
1, respectively. These maxima are relatively high, although Cu is always well below the WHO guideline value (2000 mg l
1) and no health-based guideline value exists for Zn. Most concentrations are significantly lower than the maxima. Mineral sources are likely to be Fe oxides, although desorption of these cationic species is also inhibited by the high pHs. Lithium, Rb and Cs concentrations reach up to 150, 32 and 0.22 mg l
1 respectively. The concentrations are relatively high and reflect their likely abundance in the

273

volcanic material of the host aquifer. Principal sources are likely to be micas, feldspars (Rb) and clay minerals. Strontium reaches high concentrations up to 13 mg l
1. Groundwaters are undersaturated with respect to celestite. Barium concentrations are relatively low, with an observed range of 5–260 mg l
1. Concentrations are limited by barite solubility. Many of the groundwaters are saturated or oversaturated with respect to barite. Iodine ranges between 17 and 730 mg l
1 and concentrations are relatively high compared to most groundwaters. The sources potentially include carbonate minerals in the aquifer or organic matter in the surface soil horizons. Bromide concentrations are also high in most groundwaters, varying between 0.06 and 12 mg l
1. Concentrations are highest in the more saline samples and Br correlates well with Cl. The increases are therefore likely to be caused by evaporation.

8. Isotopic compositions 8.1. d18O and d2H Observed ratios of 18O in the groundwaters range between
6.4 and
3.8% and 2H between
36 to
23%. The compositions do not show any distinct regional or depth trend, but as with the chemical data, display much variability on a local scale. The compositions show a very slight enrichment with increasing salinity but this is not marked. Ratios of 18O and 2H are
4.4 and
27%, respectively in the most saline sample collected (TDS 11,500 mg l
1). This is not the most enriched composition and is not significantly more enriched than the mean values for these isotopes. The two surface-water samples (Santa Rita pond and Tamagnoni lake) had extreme isotopic compositions of
7.1 and
2.3%, respectively for 18O and of
41 and
16%, respectively for d2H. In the case of Santa Rita, the depleted values (most negative) are considered to be representative of modern winter recharge without significant modification. In the case of the Tamagnoni lake, the more enriched composition is likely to have been caused by surface evaporation. Fig. 9 shows the trend for 18O and 2H for the Pampean groundwaters, together with the best-fit line for the data. The trend is also given for rainfall from Buenos Aires meteorological station (IAEA, 1999) which has a comparable trend with the world meteoric water line (Craig, 1961) but probably represents local rainfall compositions more closely. The best-fit line for the groundwater data has a shallower gradient than that for Buenos Aires rainfall. The enriched composition of the Tamagnoni lake sample, therefore, looks to be offset to a higher 18O value than the curve for meteoric water, which supports the suggestion that the water has undergone evaporation.

274

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

Table 5. Most of the samples investigated appear to be modern in terms of 14C model ages (negative model ages obtained using the Fontes and Garnier model). Some contain 3H and so are likely to include a component of post-1950s recharge (Table 5). However, many samples have no detectable 3H. These represent dominantly pre1950s recharge, although they are still modern in 14C terms (order of decades likely). The only sample with a registrable 14C model age was from Conhelo 4 (Table 5). Sensitivity analysis (varying the activity and 13C of soil CO2 and 13C of solid carbonate) produces a range of model ages for this sample from modern up to 1300 a. This was the deepest sample analysed (well depth 140 m) and supports the suggestion that the deeper groundwaters have slightly longer residence times. However, the sample also has measurable 3H (1.9 TU) and hence contains at least a minor component of modern recharge. Radiocarbon and 3H results indicate that the shallow groundwaters sampled are dominantly modern with short aquifer residence times. Samples from deeper parts of the aquifer have low or no detectable 3H and have probable residence times of some tens to hundreds of years. The build-up of As and other oxyanions in the

Fig. 9. Variation of 18O with 2H in the Pampean groundwaters. The best-fit regression line is also plotted for the data, alongside data for rainfall from Buenos Aires (BA) station (IAEA, 1999). This is close to the curve for the world meteoric water line (Craig, 1961).

8.2. Radiocarbon and tritium Radiocarbon activities for analysed samples range between 26 and 71 pmc (Table 4). Groundwater dating of the samples was attempted using the Fontes and Garnier (1979) model, using input parameters given in

Table 4 Exceedances of various chemical constituents above WHO guideline values (GVs; health-based) and aesthetic recommendations Determinand

WHO guideline value (GV)

Naa Cla SO4a NO3–N NO2–N NH4–Na AsT Ala Fea Mn Ba Bb F Se Cr Nib Cu Zna Mo Cd Sb Ba Pb Ub

200 250 250 11.3 0.91 1.24 10 (P)c 200 0.3 500 700 500 (P) 1.5 10 50 (P) 20 (P) 2000 (P) 3000 70 3 5 (P) 700 10 2 (P)

a b c

WHO guideline on aesthetic grounds. WHO (1998) addendum to 1993 guidelines. P: provisional value.

Units mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1 mg l
1

Number of samples

Number exceeding GV or recommendation

% Exceedance

108 108 108 108 108 83 108 108 108 108 108 108 108 34 107 107 107 107 107 107 107 107 107 107

105 47 61 51 0 0 103 6 12 0 0 107 90 11 0 0 0 0 42 0 0 0 1 107

97 44 56 47 0 0 95 5.6 11 0 0 99 83 32 0 0 0 0 39 0 0 0 1 100

275

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284 Table 5 Isotopic results for groundwater samples from La Pampaa Sample

Locality

Well depth m

970686 970690 970691 970692 970694 970697 970776 970781 970788 970795 970797

Santa Rita Epifanio Conhelo No. 4 Establ. El Tigre Estancia La Maruja Eduardo Castex No. 1 Arata No. 2 Caleufu No. 2 Trenel No. 2 Establ. Las Margaritas La Maruja No. 2

30 49 140 106 95 87 34 42 30 130 94

Water level m 9.4

74

21 28 9.4 121

18O %

2H %


5.3
6.1
4.7
4.7
5.0
4.7
4.8
4.9
5.1
4.8
6.4


26
35
32
27
31
25
24
28
27
36
35

13C %


9.1
9.8
9.1
12.2
8.9
6.2
5.9
6.9
12.3
9.1

34S %

Tritium TU

8.5

8.7

7.9 7.5

1.9 2.5 <0.5 <0.5 <0.5 0.6 1.8 <0.5 5.8

7.9 7.7 7.8 7.1

14

C pmc

14

C date years

63.1 47.4 26.4 53.0

Modern Modern Up to 1300 Modern

58.5 60.0 70.6 58.5 53.3 38.3

Modern Modern Modern Modern Modern Modern

a 14 C model dates calculated taking activity of soil CO2 as 103.7% (Fontes and Garnier, 1979), d13C of soil CO2 as
23% (Clark and Fritz, 1997), d13C of solid CaCO3 as varying between 0% and
3.7% from literature carbonate values and measured values for calcrete, activity of solid CaCO3 as 0% (ie. assuming in the CO3 aquifer to contain only dead C) and the isotopic fractionation factor between gaseous CO2 and solid CaCO3, E, calculated at aquifer temperature from values given by Deines et al. (1974). Half life of 14C is taken as 5730 a.

shallow groundwaters of the region may therefore have been a relatively rapid process. 13

8.3. d C The observed range for 13C is between
12.8 and
4.0%. Assuming that soil zone CO2 has a 13C value in the range
23 to
25% (Fontes and Garnier, 1979) and carbonate minerals in the aquifer have values close to 0%, evolution of groundwaters by carbonate reaction should drive the d13C isotopic compositions towards more enriched compositions. Indeed, 3 samples of calcrete from the aquifer gave 13C values of
1.1,
1.4 and
3.7%. The most depleted groundwater samples analysed (
12%) have undergone at least some carbonate reaction while the most enriched samples have apparently undergone considerable reaction. Observed 13C values show a weak positive correlation with HCO3 (not shown) in support of the evidence for evolution of 13C by carbonate reaction. The 13C compositions vary with well depth, with most enriched compositions being found in the shallowest groundwaters. This is probably because layers of calcrete are common in the top few metres of the aquifer in many parts and because relatively high pCO2 values can be achieved at shallow depths from bacterial reactions in the soil and are able to drive carbonate dissolution. The groundwater sample with lowest 14C activity and likely greatest aquifer residence time (Conhelo 4) has a relatively depleted 13C ratio of
9.1% (Table 4). This suggests that carbonate reaction has not been extensive in this sample and that 13C cannot be used as a residence-time indicator, but merely as a measure of degree of carbonate reaction.

8.4. d34S Data for 34S in 19 groundwater samples are plotted against SO4 concentration in Fig. 10 and given for some samples in Tables 1 and 5. The values average 7.8% and show little variation (6.3–8.7%), even given the large range of SO4 (concentrations up to 3100 mg l
1). Sulphur-isotopic compositions were measured to investigate the possibility that dissolved SO4 (and As) were derived by oxidation of pyrite in the aquifer. Pyrite has not been identified in thin sections of the loess sediments but may feasibly be present as finely-disseminated grains. Sulphur-isotopic compositions of pyrite are highly variable but generally recognised to be relatively depleted (typical range
20–0%; Clark and Fritz, 1997). The distinct range of 34S values in the Pampean groundwaters indicates that pyrite is unlikely to be

Fig. 10. Variation of 34S with SO4 in groundwaters from La Pampa.

276

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

involved in the production of SO4 in the groundwaters and is hence also not likely to be a feasible source of As. Evaporite minerals are the most likely alternative source of SO4 in the groundwaters; salt encrustations are common around sites of evaporation (depressions) during dry seasons and some of the porewaters investigated are saturated with respect to gypsum. Sulphurisotopic ratios were not measured in sulphate minerals from the study area. However, the range found for 34S in terrestrial evaporites, although large, (
15%–+10%; Clark and Fritz, 1997), covers the range of values observed for SO4 in the groundwaters. Little S-isotopic fractionation is involved in dissolution or precipitation of gypsum (Clark and Fritz, 1997).

9. Porewater chemistry Chemical profiles for porewaters from the two cored boreholes at Talleres Norte and Tamagnoni are shown in Figs. 11 and 12 respectively. All porewater samples are from the saturated zone. As mentioned in Section 6, careful attempts were made during drilling to minimise the effects of cross-contamination of porewaters by groundwater from overlying horizons, although lack of contamination cannot be guaranteed in all samples, especially some from the shallowest depths where calcrete layers were present. Nonetheless, the contamination is thought to be minor and not materially important to the overall interpretation of the chemical trends. 9.1. Talleres Norte borehole As the porewaters are from the saturated zone of the aquifer, analysis of moisture contents gives an indication of the porosity of the sediments. For the Talleres Norte samples, the moisture contents are variable between 22 and 40% (dry weight) but average around 30% (equivalent to ca. 45% porosity) throughout the profile down to its base at 26.5 m depth. Salinity is relatively low, with SEC values varying between around 600 and 2000 mS cm
1 (Fig. 11). Salinity, indicated by SEC, Na, Cl, SO4 and B for example, increases gradually with depth, although the topmost sample also has relatively high salinity. The porewater compositions compare well with those of pumped groundwaters from the site, which reveal increasing salinity with depth: SEC values of 1200 mS cm
1 were found at 12 m depth and 6800 mS cm
1 at 48 m, below the base of the cored profile. Porewater pH shows little variation but, like the pumped groundwaters, is mostly high. Values are slightly higher (>8.5) between 12 and 25 m than at shallower levels. Calcium and Mg also vary little, although the topmost sample from close to the water table has much

higher concentrations of these elements (Fig. 11). As with the pumped groundwaters, concentrations of these are low relative to Na. Nitrate-N decreases with depth, with a maximum of 9 mg l
1 close to the water table, decreasing down to < 0.2 mg l
1 from 16 m depth to the base of the borehole. The concentration in the pumped groundwaters from Talleres Norte was 2.9 mg l
1 at 48 m depth. Although this concentration is relatively low, the fact that it is higher than in the deepest part of the cored borehole suggests that a degree of drawdown of higher NO3 water from near the water table has occurred in the pumped borehole. The low NO3–N concentrations of deeper porewaters are not thought to relate to NO3–N reduction as the porewaters are largely oxidising. It is more likely that the low values reflect lack of pollution at depths greater than 16 m and historically efficient capture of atmospheric N by soils and plants. The presence of highest NO3–N concentrations at shallow depths mirrors the situation in the pumped groundwaters described above (Fig. 3). Concentrations of Fe and Mn (and Al, not shown) are low in many of the Talleres Norte porewater samples but are relatively high in some samples from the middle section. The 3 elements are correlated which suggests that the affected samples have a significant colloidal component ( <0.45 mm). Concentrations of Fe, Mn and Al in the pumped groundwaters from the site were found to be around 500, 7 and 900 mg l
1, respectively. The Fe, Mn and Al concentrations in porewaters from the middle section are much higher, reaching up to 3.9 mg l
1, 89 mg l
1 and 3.5 mg l
1, respectively. Concentrations of some of the trace elements, especially the REE, Y, Th and Co, show similarly spiky patterns which are probably due to association with colloidal material rather than the dissolved component of the porewaters. Under the alkaline conditions of the porewaters, mobilisation of these cationic species is not expected. The patterns for the REE, Y, Th and Co differ from those for As and the other oxyanions and so colloidal association is not thought to be the main control on the profiles of these latter elements. Fluoride varies between 2.9 and 25.7 mg l
1 with the lowest values observed at around 9–10 m depth and highest values present from around 17 m depth to the base of the hole. The porewaters are saturated with respect to fluorite in all samples except those with lowest F concentrations, around the 10 m interval. Saturation with respect to fluorite ensures that concentrations of F are limited at around the 26 mg l
1 maximum. Concentrations of F in the pumped Talleres Norte groundwater samples are broadly comparable at 12 m depth (15 mg l
1) but appear to be lower below the depth range of the cored borehole (6.8 mg l
1 at 48 m depth). The V profile is similar to that of F, with a maximum of 1.7 mg l
1 at 17.5 m depth. The As profile also shows

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

277

Fig. 11. Chemical profiles for porewaters extracted from sediment cores from Talleres Norte borehole, Eduardo Castex [35 54.30 S 64 18.070 W]. MC: moisture content (% dry weight).

278

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

Fig. 12. Chemical profiles for porewaters extracted from sediment cores from Tamagnoni borehole, 10 km south of Eduardo Castex [35 59.550 S 64 15.570 W]. MC: moisture content (% dry weight).

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

some similarities with V and F, with a corresponding minimum at around 9–10 m depth and a maximum of 530 mg l
1 at 21 m. The profiles of As, V and F all show some correlation with HCO3, as with pumped groundwater samples from the regional survey. The concentrations also tend to be highest where pH values are above 8.5, although the correlation with pH is not so striking in the porewaters as in the pumped groundwater samples. Concentrations of B, Se and U have some similarities, with generally increasing trends with depth and maxima at around 25 m. Concentrations reach up to 3.4 mg l
1, 5 mg l
1 and 95 mg l
1 respectively for these elements. Molybdenum is generally < 60 mg l
1 throughout the profile but increases significantly, to >100 mg l
1 in the lowermost 2 m of the borehole. 9.2. Tamagnoni borehole As noted above, the Tamagnoni borehole is located close to a localised depression which has been water-filled over the last few years and which undergoes evaporation. The site is therefore considered to be one of seasonal discharge of slow-moving groundwater and the locus of seasonal evaporative concentration. One of the most striking features of the Tamagnoni profile is the much enhanced salinity of the porewaters in the topmost 5 m of core (Fig. 12). SEC values reach up to 10,000 mS cm
1 at the top of the profile and concentrations of most of the major elements (Na, Cl, Ca, Mg, K, SO4, NO3-N) as well as Sr, Li and Rb (not shown) are correspondingly high. The enhanced concentrations are most likely to be the result of evaporative concentration of solutes in the zone of water-table fluctuation. Bailed samples taken from 5.5, 16 and 30.5 m depth during drilling show comparable salinity variations with depth. Porewaters in the shallowest part of the borehole are saturated with respect to gypsum as a result of the enhanced Ca and SO4 concentrations. They are also supersaturated with respect to calcite and dolomite. The evaporative process has led to extremes of many elements but the concentrations of constituents detrimental to health are particularly notable. Nitrate (N) concentrations reach up to 620 mg l
1 at the top of the profile, a value 55 times the WHO guideline value for NO3–N in drinking water. Although the Tamagnoni porewaters are not samples of drinking water, the concentrations illustrate the potential detrimental effects of evaporation on water quality and potential problems arising from locating abstraction boreholes close to depressions where groundwater salinity is locally high. The concentration of NO3–N in a bailed sample from 5.5 m depth in the borehole during drilling was 310 mg l
1, although the concentration in the abstraction borehole at Tamagnoni used for potable supply was only 12 mg l
1 (though still slightly above the WHO guideline value; Table 1).

279

Porewater pH varies mainly between 7 and 8 in the topmost saline portion of the profile but increases to around 8.6 from 10 m depth to the base of the borehole. Concentrations of HCO3 show a similar trend, being less than 500 mg l
1 in the topmost part and close to 1000 mg l
1 in many samples below 10 m depth. The variations reflect the transition from Na–Cl dominant waters near surface to Na–HCO3 waters below the major zone of evaporation. Moisture contents in the Tamagnoni samples range between 26 and 40% (dry weight) with most being around 30%. Again this represents an average porosity of around 45%. As with the Talleres Norte profile, some high concentrations of Fe and Mn (and Al) occur in the porewaters at depths of 17–23 m (Fig. 12). At the pH of the groundwaters, these are probably colloidal fractions ( < 0.45 mm) and are likely to be the reason why concentrations of REE, Y, and Th are correspondingly high at these depths. Again as with Talleres Norte, the colloidal components are not believed to have a major control on the distribution of dissolved oxyanions as the profiles do not closely match. Arsenic concentrations are low in the topmost (saline) section of the profile, but increase with depth to a maximum of 7500 mg l
1 at 27 m and decrease slightly below this horizon to the base of the borehole (Fig. 12). The As profile is distinct from indicators of salinity such as Cl and clearly indicates that evaporation is not responsible for the accumulation of dissolved As. This was also apparent from the pumped groundwaters collected in the regional survey (Fig. 8). The highest As concentration is more than 10 times greater than the maximum observed in the Talleres Norte porewaters. The concentrations are also significantly higher than those found in the groundwater from the Tamagnoni abstraction borehole just a few metres away, although this was also very high at 1360 mg l
1. The high observed As concentrations may reflect in part the fact that the area is a zone of periodic groundwater discharge, where groundwater flow is sluggish and flushing from the aquifer locally has been restricted. It is also notable that the highest concentrations of porewater As occur in a zone of the profile which was particularly dry (24–28 m depth; Fig. 12). Indeed, the maximum observed concentration occurred immediately below a clayey horizon so dry that insufficient water could be extracted for chemical analysis (hence the gap at 27 m in the Tamagnoni profile; Fig. 12). This information suggests that groundwater movement is particularly restricted in this depth zone due to fine grain size. The dry clayey zone is also an area of enhanced mottling due to the presence of Mn-oxide coatings and cement. This combination of groundwater flow conditions and localised variation in mineralogy and lithology is believed to be responsible for the extremely high porewater-As

280

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

concentrations in this part of the Tamagnoni profile. Derivation of the As from the Mn oxides in the sediments may be an important process (Smedley et al., 2000). Porewater profiles for other anion and oxyanion species such as F, V, U, B, Se and Mo show some similarities with that for As in that concentrations generally increase with depth, although the profiles of each do not correspond closely. The processes of release of these elements into the porewaters are believed to be broadly similar to those for As but differences between the elements are likely to reflect differing transport and retardation properties of the elements and resultant chromatographic separation over the vertical profile with time.

10. Discussion Arsenic concentrations in the Pampean groundwaters show a large range, although most groundwater sources investigated in this study were found to be elevated: 95% above the WHO guideline value and 73% above the Argentine standard. Sample collection was close to random and hence the exceedance statistics are believed to be representative of the area as a whole. The data indicate that the groundwaters are seriously enriched with As, not only because of the percentage of samples exceeding guideline limits but also the absolute concentrations of many are high (53% exceed 100 mg l
1, 26% exceed 300 mg l
1 and 6.5% exceed 1000 mg l
1). For comparison, a recent survey of shallow groundwaters (wells < 150 m deep) from the As-affected aquifers of Bangladesh found that around 46% exceeded the WHO guideline value, while 27% exceeded the Bangladesh national standard of 50 mg l
1 (Kinniburgh et al., 2001). The distribution of exceedances in Bangladesh was also patchy however, with some areas notably more enriched than others. The lack of distinct regional trends in chemical composition of the groundwaters is likely a reflection of the general uniformity of composition of the Pampean sediments across the region. Although grain-size and texture are variable on a local scale and do show some control on As concentrations (e.g. the clayey horizons have high porewater As concentrations in Tamagnoni borehole), the Pampean sediments do not show major lithological differences across the region. Hence chemical variations are likely to be related predominantly to physical and hydrochemical factors rather than to major lithological variations. The high degree of spatial variability in groundwater chemistry over short distances and vertical variations in groundwater chemistry observed in cored boreholes suggests that groundwater movement has been restricted and the groundwaters are poorly mixed. This arises

probably as a result of restricted recharge in the semiarid climate of the region, low hydraulic gradients and limited permeability of the Pampean silts and fine sands. Investigation of sediments from the loess aquifer (Smedley et al., 2000) indicates that As concentrations in the solid phase are close to world-average values (5– 10 mg kg
1; Webster, 1999), although some spatial association exists between high As in the groundwaters and high As in the sediments. Smedley et al. (2000) found total As concentrations ranging between 3 and 18 mg kg
1, the higher concentrations tending to be from finer-grained (silty) sediments. The mineral sources of As and the other anions and oxyanions in solution are not clearly established. Potential sources include primary minerals in the loess and incorporated volcanic materials, many of which have been well weathered during aeolian transport and since deposition. Primary magnetite and ilmenite constitute potential sources of many trace elements (V and As in particular). Dissolved F may be derived by weathering of detrital biotite and apatite, and P is also likely to be derived from apatite. Some primary silicate minerals may also contribute trace elements to solution on weathering. Indeed, Nicolli et al. (1989) suggested that breakdown of volcanic glass was responsible for the high-As groundwaters in neighbouring Co´rdoba Province. However, secondary metal oxides, especially of Fe and Mn, are considered to be important and perhaps dominant sources since they are relatively abundant in the sediments and show some association with the distribution of solid-phase As (Smedley et al., 2000). Desorption of anionic species from Fe oxides in particular will inevitably occur under the ambient high-pH groundwater conditions in the Pampean aquifers. The effect has been well demonstrated for various oxyanion-forming elements on hydrous ferric oxide (Hfo) and goethite in particular (Dzombak and Morel, 1990; Manning and Goldberg, 1996; Hiemstra and Van Riemsdijk, 1999). Desorption reactions are likely to be enhanced by mutual competition between oxyanion species for available adsorption sites on the oxides present (Manning and Goldberg, 1996). Additional desorption may be promoted as a result of ageing of the Fe oxides during sediment diagenesis, with resultant changes in oxide structure and crystallinity (e.g. Zielinski et al., 1983; Smedley and Kinniburgh, in press). Desorption of oxyanions from secondary oxide minerals implies a twostage process where weathering of primary detrital minerals provides the original trace-element source. The oxidising condition of the groundwaters leads to As being present dominantly as As(V). This reflects the unconfined condition of the aquifers and the paucity of available organic matter and sulphide minerals in the sediments for consumption of dissolved O2 and reduction of other oxidised species. The presence of Mn oxides as cements and occasional nodules also probably

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

aids the maintenance of As in the oxidised form. The efficient oxidation capacity of Mn oxides for As(III) is well-documented (Oscarson et al., 1981a,b). The fact that many of the anion and oxyanion constituents present in the groundwaters have been assigned WHO drinking-water guideline values indicates that some health risk is attached to them from chronic exposure. The groundwaters are therefore potentially detrimental as potable water sources. However, since these ions are strongly correlated because of their similar hydrochemical behaviour, identification of sources with lowest As concentrations will coincidentally identify sources with the lowest concentrations of the other problem elements in this group. It is unfortunate however that the lowest As concentrations tend to be found in groundwaters which are more saline and hence also largely unsuitable for potable use. Of the samples investigated with As concentrations <50 mg l
1, only 9 (32%) have SEC concentrations less than 2000 mS cm
1 and hence are of low salinity. These are also Na-HCO3dominated waters but have pH values close to neutral. Overall groundwater quality is therefore poor and hence mitigation measures are needed for provision of potable water. While some benefit may be gained from siting new boreholes away from groundwater discharge zones and from deepening boreholes, this is likely to be limited because of the poor observed quality of most groundwaters. Mitigation measures require groundwater treatment or an additional possibility exists from the collection and storage of rainwater. Saline and high-pH groundwaters similar to those observed in La Pampa are typical features of arid and semi-arid regions. In such conditions, evaporative concentration of solutes may be an important process in the near-surface and limited recharge means that water– rock reactions can proceed to a relatively advanced stage without significant dilution. Analogue situations to that of La Pampa occur in parts of the south-west USA, Mexico and Chile. One good example is the arid region of the San Joaquin Valley, California. In the Tulare Basin, a closed basin in the southern part of the Valley, shallow groundwaters (< 7 m) commonly have high concentrations of major ions and many trace elements as a result of evaporative concentration (Fujii and Swain, 1995). The groundwaters often have high As concentrations, commonly under oxidising conditions where As(V) is the dominant dissolved species, and many have associated high concentrations of B, Mo and U as well as Se. These groundwaters have high pH values and this is taken to be an important criterion for generating high concentrations of anions in solution. Processes controlling groundwater compositions in the Tulare Basin are believed to be very similar to those in La Pampa. Reducing conditions are also present in some parts of the aquifers, and these also experience As problems. In

281

these however, the concentrations of Se, U and Mo are usually low (Fujii and Swain, 1995). In northern Mexico, similar geochemical conditions occur in the arid region of Lagunera. Here, groundwaters are predominantly oxidising with neutral to high pH (6.3–8.9; Del Razo et al., 1990). High As concentrations are common and dominated by As(V) and the groundwaters have correspondingly high F concentrations. In northern Chile, both surface and groundwaters from the arid region around Antofagasta have high As concentrations again dominated by As(V), high salinity (largely caused by evaporation) and high B concentrations (Ca´ceres et al., 1992). These intercomparisons mean that the hydrochemical features observed in La Pampa are representative of many other arid and semi-arid areas, particularly those where recent loess deposits exist. The conclusions reached therefore have wider relevance to As and other oxyanion problems in aquifers elsewhere.

11. Conclusions Groundwaters from northern La Pampa Province have a range of inorganic-quality problems. The most serious include high salinity and high concentrations of As, F, B, Mo, U and Se in particular. The high salinity values are generated by evaporation and this may also be partially responsible for accumulation of some trace elements. Selenium shows a weak positive correlation with salinity indicators, for example. However, evaporation is not the main process responsible for the accumulation of dissolved As and many of the other anions and oxyanions. Arsenic in the groundwaters is positively correlated with pH as well as with V and F in particular. High concentrations of these in pumped groundwaters and porewaters do not correspond with salinity indicators such as Cl and preclude evaporation as a dominant influence on their aqueous distribution. Many of the groundwaters have high pH (up to 8.7) and alkalinity (HCO3 up to 1440 mg l
1). The high pH values are generated by carbonate and silicate reaction (e.g. weathering of albite and volcanic glass). The groundwaters are oxidising with high concentrations of dissolved O2 and NO3–N and low Fe, Mn, NH4–N and NO2–N concentrations. Arsenic occurs in solution mainly as As(V). Dissolved As correlates positively with pH and alkalinity. Desorption of As and the other oxyanions from metal oxides, particularly of Fe and Mn, under the high-pH groundwater conditions is considered the most important control on their aqueous concentrations. Secondary Fe and Mn oxides are abundant in the sediments and primary ilmenite and magnetite are also present. Dissolved As may also be derived at least in part from

282

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

weathering of primary silicate minerals (e.g. biotite, volcanic glass) in the loess and ash. Fluoride and P are likely to derive additionally from detrital apatite. Relatively high concentrations of As (and F, V, B, Mo, U) tend to be found in groundwaters pumped from close to the water table. The mobilisation of the trace anions and oxyanions appears to be greatest where chemical reactions (carbonate, silicate) are occurring most actively. High concentrations of As and other anions and oxyanions also tend to be present in groundwaters from localised depressions which may become intermittent discharge zones. Such depressions are likely to be zones of restricted groundwater movement where flow is directed inwards towards the centre of the depression, although it is diluted seasonally by surface runoff which also collects in the depression. Fine-grained sediments may also be preferentially concentrated in the depressions due to aeolian reworking and may further restrict localised groundwater movement in these zones. Restricted flow without significant flushing is likely to be one reason for the accumulation of As and the other problem elements in solution in these areas. Shallow groundwaters in these localised discharge zones also have increased salinity as a result of evaporation. Saturated-zone porewaters from a cored borehole adjacent to one of the depressions (Tamagnoni) demonstrated increased salinity in the topmost 3 m as a result of evaporation at the surface. These porewaters were supersaturated with calcite and dolomite and saturated with gypsum. At these shallow levels, concentrations of As and the other oxyanions were relatively low but at greater depth (25–30 m depth), As reached extremes up to 7500 mg l
1 and V up to 12 mg l
1, with correspondingly high concentrations of many of the other oxyanion species. Accumulation at high pH under locally restricted groundwater flow conditions is considered responsible. The semi-arid conditions of the Pampean region, low recharge totals, low sediment permeability and limited topographic variation suggest that groundwater flow in the Pampean aquifer is generally slow. This is supported by the high degree of short-range spatial variability in groundwater chemistry which highlights lack of mixing. Low 3H concentrations of many groundwaters indicate that these comprise dominantly pre-1950s recharge, although the lack of observed groundwaters with old 14 C model ages suggests that there is at least a degree of active recharge which introduces modern radiocarbon (and in some, 3H) to the aquifer. One deep groundwater sample (140 m depth) had the lowest observed 14C value (26.4 pmc) and gives a model age up to 1300 a BP. Older groundwaters therefore occur at greater depth, although there is no evidence for the presence of very old groundwaters in the aquifer. Modern recharge is being introduced at shallow depths in the aquifer as evidenced by detectable 3H in shallow groundwater samples. Introduction of modern recharge is also indicated by

rising water levels over the last few decades in response to increases in rainfall. Comparisons of groundwater chemistry in La Pampa with waters in northern Mexico, Chile and south-west USA suggest that the conclusions found for La Pampa may be applicable elsewhere in determining regions vulnerable to As and related water-quality problems. It is concluded that, under oxidising conditions, vulnerable aquifers potentially occur where several important criteria coincide: semi-arid climatic conditions with limited recharge where high-pH groundwaters can be generated; young (Quaternary) sediments (or volcanogenic sediments); and slow groundwater-flow conditions. Such aquifers are likely to have been poorly flushed over the geologically-short timescale since deposition and hence will have had little opportunity for removal of trace elements such as As from the aquifer. These oxidising, high-pH groundwaters are distinct from those developing As problems in strongly reducing conditions such as in the Bengal Basin, Inner Mongolia and southern Hungary. However, in these too, slow groundwater flow and poor aquifer flushing are doubtless contributors to the high dissolved As concentrations.

Acknowledgements We are especially grateful to staff of Direccio´n de Aguas, Santa Rosa, La Pampa Province, for providing assistance with groundwater sampling and provision of hydrogeological data. Chemical and stable-isotopic analysis of water and sediment samples was carried out by staff at the British Geological Survey, Wallingford, UK. Sulphur-isotopic analysis was carried out in part by the NERC Isotope Geology Laboratory, Keyworth, UK and part by Geochron Laboratories, Cambridge, Massachusetts, USA. Radiocarbon and 3H analysis was carried out by Instituto de Geocronologı´a y Geologı´a Isoto´pica (INGEIS), Buenos Aires, Argentina. We are also grateful to A.H. Welch and R.B. Wanty for providing thorough and constructive reviews, and to D.G. Kinniburgh for reviewing an earlier version of the manuscript. Research was carried out with funding provided by UK-DFID under its Knowledge and Research (KAR) programme, project number R6491. INGEOQUI has participated with additional funding from PIP No. 4430, CONICET, Argentina. PLS and DMJM publish with the permission of the Director, British Geological Survey.

References Argu¨ello, R.A., Cenget, D.D., Tello, E.E., 1938. Cancer y arsenicismo regional ende´mico en Co´rdoba. Argent. Rev. Dermatosyphilol. 22, 461–487.

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284 Ayerza, A., 1918. rsenicismo regional ende´mico (keratodermia y melanodermia combinadas). Bol. Acad. Nac. Medicina, Buenos Aires, Argentina I, 11–41. Bannert, D., 1974. Hydrogeological Investigations in the Pampa of Argentina. Federal Geological Survey, SR No 330, Federal Republic of Germany. Besuschio, S.C., Desanzo, A.C., Perez, A., Croci, M., 1980. Epidemiological associations between arsenic and cancer in Argentina. Biol. Trace Element Res. 2, 41–55. Biagini, R., Rivero, M., Salvador, M., Co´rdoba, S., 1978. Hidroarse´nicismo cro´nico y cancer de pulmo´n. Arch. Argent. Dermatol. 48, 151–158. Blarasin, M., Cabrera, A. Matteoda, E. Villalba, Y.G., 2000. Condicionamientos Hidrogeolo´gicos Sobre la Presencia de Arsenico en Aguas Subterra´neas en el Marco de los Geoindicadores en el sur de Co´rdoba. Report of the University of Rı´o Cuarto, Co´rdoba, Argentina. Ca´ceres, L., Gruttner, V.E., Contreras, R., 1992. Water recycling in arid regions: Chilean case. Ambio 21, 138–144. Chen, C.J., Chuang, Y.C., Lin, T.M., Wu, H.Y., 1985. Malignant neoplasms among residents of a blackfoot diseaseendemic area in Taiwan: high-arsenic artesian well water and cancers. Cancer Res. 45, 5895–5899. Cı´rculo Me´dico del Rosario, 1917. Sobre la nueva enfermedad descubierta en Bell-Ville. Rev. Me´dica del Rosario, Rosario, Argentina VII, 485. Clapperton, C., 1983. The glaciation of the Andes. Quat. Sci. Rev. 2, 83–155. Clark, I., Fritz, P., 1997. Environmental Isotopes in Hydrogeology. Lewis Publishers. Craig, H., 1961. Isotopic variations in meteoric waters. Science 133, 1702–1703. Cullen, W.R., Reimer, K.J., 1989. Arsenic speciation in the environment. Chem. Reviews 89, 713–764. Das, D., Chatterjee, A., Mandal, B.K., Samanta, G., Chakraborti, D., Chanda, B., 1995. Arsenic in groundwater in six districts of West Bengal, India: the biggest arsenic calamity in the world. Part 2. Arsenic concentration in drinking water, hair, nails, urine, skin-scale and liver tissue (biopsy) of the affected people. Analyst 120, 917–924. Deines, P., Langmuir, D., Harmon, R.S., 1974. Stable carbon isotope ratios and the existence of a gas phase in the evolution of carbonate ground waters. Geochim. Cosmochim. Acta 38, 1147–1163. Del Razo, L.M., Arellano, M.A., Cebria´n, M.E., 1990. The oxidation states of arsenic in well-water from a chronic arsenicism area of northern Mexico. Environ. Pollution 64, 143–153. Deuel, L.E., Swoboda, A.R., 1972. Arsenic solubility in a reduced environment. Soil Sci. Soc. Amer. Proc. 36, 276–278. Driehaus, W., Jekel, M., 1992. Determination of As(III) and total inorganic arsenic by on-line pretreatment in hydride generation atomic absorption spectrometry. Fres. J. Anal. Chem. 343, 352–356. Dzombak, D.A., Morel, F.M.M., 1990. Surface Complexation Modeling: Hydrous Ferric Oxide. John Wiley & Sons, New York. Elorriaga, E.E., Tullio, J.O., 2000. Estructuras del subsuelo y su influenca en la morfologı´a en el norte de la Provincia de La Pampa. Actas X Congreso Latinamericano de Geologı´a y VI Congreso Nacional de Geologı´a Econo´mica III, 499–506.

283

Fontes, J.-C., Garnier, J.-M., 1979. Determination of the initial 14C activity of the total dissolved carbon: a review of the existing models and a new approach. Water Resour. Res. 15, 399–413. Freeze, R.A., Cherry, J.A., 1979. Groundwater. Prentice Hall. Fujii, R., Swain, W.C., 1995. Areal Distribution of Selected Trace Elements, Salinity, and Major Ions in Shallow Ground Water, Tulare Basin, Southern San Joaquin Valley, California. USGS Water-Resources Investigations Report 954048. Fujii, R., Deverel, S.J., Hatfield, D.B., 1988. Distribution of selenium in soils of agricultural fields, western San Joaquin Valley, California. Soil Sci. Soc. Amer. J. 52, 1274–1283. Hem, J.D., 1992. Study and interpretation of the chemical characteristics of natural water. USGS Water-Supply Paper 2254. Herna´ndez, M.A., Gonza´lez, N., 1999. Impact of rising piezometric levels on Greater Buenos Aires due to partial changing of water services infrastructure. In: Chilton, P.J. (Ed.), Proc. XXVII IAH Congress on Groundwater in the Urban Environment, Nottingham, UK, 21–27 September, Vol. 1: Problems, Processes and Management. Hiemstra, T., Van Riemsdijk, W.H., 1999. Surface structural ion adsorption modeling of competitive binding of oxyanions by metal (hydr)oxides. J. Colloid Interf. Sci. 210, 182–193. Hopenhayn-Rich, C., Biggs, M.L., Fuchs, A., Bergoglio, R., Tello, E.E., Nicolli, H., Smith, A.H., 1996. Bladder cancer mortality associated with arsenic in drinking water in Argentina. Epidemiol. 7, 117–124. IAEA, 1999. Global Network for Isotopes in Precipitation Programme, International Atomic Energy Agency, Vienna and World Meteorological Office, http://www.iaea.org/programs/ri/qnip. Kinniburgh, D.G., Smedley, P.L., Trafford, J.M., Milne, C.J. Huq, S.M.I., Ahmed, K.M., Burden, S., 2001. The National Hydrochemical Survey. In: Kinniburgh, D.G., Smedley, P.L. (Eds.), Arsenic Contamination of Groundwater in Bangladesh. Volume 2: Final Report. British Geological Survey Report, WC/00/19. British Geological Survey, UK (Chapter 6), pp. 77–103. Korte, N.E., Fernando, Q., 1991. A review of arsenic(III) in groundwater. Critical Reviews in Environ. Control 21, 1–39. Kro¨hling, D.M., 1999. Upper Quaternary geology of the lower Carcaran˜a´ Basin, North Pampa. Argentina. Quat. Internl. 57/58, l135–148. Logan, W.S., Rudolph, D.L., 1997. Microdepression-focused recharge in a coastal wetland, La Plata, Argentina. J. Hydrol. 194, 221–238. Manning, B.A., Goldberg, S., 1996. Modeling competitive adsorption of arsenate with phosphate and molybdate on oxide minerals. Soil Sci. Soc. Amer. J. 60, 121–131. Michelena, R.O., Irurtia, C.B., 1995. Susceptibility of soil to wind erosion in La Pampa Province, Argentina. Arid Soil Res. Rehabil. 9, 227–234. Middlemost, E.A.K., 1985. Magmas and Magmatic Rocks. Longman, London. Nicolli, H.B., Merino, M.H., in press. High contents of F, As and Se in groundwater of the Carcaran˜a´ river basin, Argentine Pampean Plain. Environ. Geol. Nicolli, H.B., Suriano, J.M., Go´mez Peral, M.A., Ferpozzi, L.H., Baleani, O.H., 1989. Groundwater contamination with

284

P.L. Smedley et al. / Applied Geochemistry 17 (2002) 259–284

arsenic and other trace-elements in an area of the Pampa, province of Co´rdoba, Argentina. Environ. Geol. Water Sci. 14, 3–16. Nordstrom, D.K., Alpers, C.N., Ptacek, C.J., Blowes, D.W., 2000. Negative pH and extremely acidic mine waters from Iron Mountain California. Environ. Sci. Technol. 34, 254–258. O’Brien, R., Keller, C.K., Smith, J.L., 1996. Multiple tracers of shallow ground-water flow and recharge in hilly loess. Groundwater 34 (4), 675–682. Oscarson, D.W., Huang, P.M., Defosse, D., Herbillion, A., 1981a. Oxidative power of Mn(IV) and Fe(III) oxides with respect to As(III) in terrestrial and aquatic environments. Nature 291, 50–51. Oscarson, D.W., Huang, P.M., Liaw, W.K., 1981b. Role of manganese in the oxidation of arsenite by freshwater lake sediments. Clays Clay Min. 29, 219–225. Pearce, J., 1998. A Petrographic Study of Potential Arsenic Sources in Loess Deposits from the Pampa Region, Argentina. British Geological Survey Technical Report, WG/98/34R. Rabassa, J., De Francesco, F., 1986. The loess of central and western Buenos Aires Pampas, Argentina. Loess Lett (Leicester, Waterloo) 16, 36–37. Schreiber, M.E., Simo, J.A., Freiberg, P.G., 2000. Stratigraphic and geochemical controls on naturally occurring arsenic in groundwater, eastern Wisconsin, USA. Hydrogeol. J. 8, 161– 176. Smedley, P.L., Kinniburgh, D.G., in press. Source and behaviour of arsenic in natural waters. In: UN Synthesis Report on Arsenic, WHO, Geneva (Chapter 1). Smedley, P.L., Macdonald, D.M.J.M., Nicolli, H.B., Barros, A.J., Tullio, J.O., Pearce, J.M., 2000. Arsenic and Other Quality Problems in Groundwater from Northern La Pampa Province, Argentina. British Geological Survey Technical Report, WC/99/36. Smith, A., Goycolea, M., Haque, R., Biggs, M.L., 1998. Marked increase in bladder and lung cancer mortality in a

region of Northern Chile due to arsenic in drinking water. Am. J. Epidemiol. 147, 660–669. Smith, A., Hopenhayn-Rich, C., Bates, M., Goeden, H., HertzPicciotto, I., Duggan, H., Wood, R., Kosnett, M., Smith, M., 1992. Cancer risks from arsenic in drinking water. Environ. Health Persp. 97, 259–267. Teruggi, M.E., 1957. The nature and origin of Argentine loess. J. Sed. Petrol. 27, 322–332. Tonni, E.P., Cione, A.L., Figini, A.J., 1999. Predominance of arid climates indicated by mammals in the pampas of Argentina during the Late Pleistocene and Holocene. Palaeogeog. Palaeoclimatol. Palaeoecol. 147, 257–281. Trafford, J.M., 1986. Investigations into the storage of groundwater containing trace levels of arsenic, prior to analysis by hydride generation combined with ICP–AES. British Geological Survey Report 86/8. Tseng, W.-P., Chu, H.-M., How, S-H., Fong, J-M., Lin, C-S., Yeh, S., 1968. Prevalence of skin cancer in an endemic area of chronic arsenicism in Taiwan. J. Nat. Cancer Inst. 40, 453–463. Wanty, R.B., Goldhaber, M.B., 1992. Thermodynamics and kinetics of reactions involving vanadium in natural systems: accumulation of vanadium in sedimentary rocks. Geochim. Cosmochim. Acta 56, 1471–1483. Webster, J.G., 1999. Arsenic. In: Marshall, C.P., Fairbridge, R.W. (Eds.), Encyclopaedia of Geochemistry. Chapman and Hall, London, pp. 21–22. WHO, 1998. Guidelines for Drinking-Water Quality, 2nd Edition. World Health Organization, Geneva. Addendum to Volume 2. Za´rate, M.A., Fasano, J.L., 1989. The Plio-Pleistocene record of the central Eastern Pampas, Buenos-Aires Province, Argentina — the Chapadmalal case-study. Palaeogeog. Palaeoclimatol. Palaeoecol. 72, 27–52. Zielinski, R.A., Bloch, S., Walker, T.R., 1983. The mobility and distribution of heavy metals during the formation of first cycle red beds. Econ. Geol. 78, 1574–1589.