Journal of Contaminant Hydrology 99 (2008) 8–21

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology

j ourna l homepage: www.e lsev ie r.com/ locate / jconhyd

Arsenic in groundwaters in the Northern Appalachian Mountain belt: Areview of patterns and processes

Stephen C. PetersLehigh University, Department of Earth and Environmental Sciences, Bethlehem, PA, USA

a r t i c l e i n f o

E-mail address: scp2@lehigh.edu.

0169-7722/$ – see front matter © 2008 Elsevier B.V.doi:10.1016/j.jconhyd.2008.04.001

a b s t r a c t

Article history:Received 14 August 2007Received in revised form 1 April 2008Accepted 4 April 2008Available online 20 June 2008

Naturally occurring arsenic in the bedrock of the Northern Appalachian Mountain belt wasfirst recognized in the late 19th century. The knowledge of the behavior of arsenic ingroundwater in this region has lagged behind nearly a century, with the popular pressreporting on local studies in the early 1980s, and most peer-reviewed research articles onregional patterns conducted and written in the late 1990s and early 2000s. Research reportshave shown that within this high arsenic region, between 6% and 22% of households usingprivate drinking water wells contain arsenic in excess of 10 µg/L, the United StatesEnvironmental Protection Agency's maximum contaminant level. In nearly all reports,arsenic in drinking water was derived from naturally occurring geologic sources, typicallyarsenopyrite, substituted sulfides such as arsenian pyrite, and nanoscale minerals such aswesterveldite. In most studies, arsenic concentrations in groundwater were controlled by pHdependent adsorption to mineral surfaces, most commonly iron oxide minerals. In somecases, reductive dissolution of iron minerals has been shown to increase arsenicconcentrations in groundwater, more commonly associated with anthropogenic activitiessuch as landfills. Evidence of nitrate reduction promoting the presence of arsenic(V) and iron(III) minerals in anoxic environments has been shown to occur in surface waters, and in thismanuscript we show this process perhaps applies to groundwater. The geologic explanationfor the high arsenic region in the Northern Appalachian Mountain belt is most likely thecrustal recycling of arsenic as an incompatible element during tectonic activity. Accretion ofmultiple terranes, in particular Avalonia and the Central Maine Terrane of New Englandappear to be connected to the presence of high concentrations of arsenic. Continued tectonicactivity and recycling of these older terranes may also be responsible for the high arsenicobserved in the Triassic rift basins, e.g. the Newark Basin. There are only two well-knowncases of anthropogenic contamination of the environment in the northern AppalachianMountain belt, both of which are industrial sites with surface contamination at thatinfiltrated the local groundwater.

© 2008 Elsevier B.V. All rights reserved.

Keywords:ArsenicGeologyGroundwaterGeochemistry

1. Introduction

The knowledge and understanding of arsenic behavior ingroundwaters of the northern half of the AppalachianMountain belt grew tremendously between the years 1999and 2007. Researchers completed dozens of studies and

All rights reserved.

analyzed thousands of water, rock, and soil samples. Technol-ogy improved the detection limit for arsenic measurement innatural waters by several orders of magnitude (e.g. Klaue andBlum, 1999), and public interest was heightened by aregulatory decision setting a new United States Environmen-tal Protection Agency maximum contaminant level. Thearsenic problem as outlined here has been dwarfed in recentyears by large research initiatives in Bangladesh, Taiwan, andother countries where the impacts to human health are acute

9S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

and tragic. The health effects of arsenic to the population inthe Northern Appalachian Mountain belt are most certainlynot of the same magnitude, though the long-term exposureeffects are unknown and the geologic origins and regionalmobilization mechanisms are worthy of study.

Arsenic occurs as a trace element in many geologicmaterials, and as such, is also found in many natural waters.The detailed biogeochemical properties and behavior ofarsenic in natural waters have been reviewed by others andare not repeated here (e.g. Welch et al., 2000; Smedley andKinniburgh, 2002). Most of the articles discussed here werereleased after the publication of these reviews, and are notcovered in depth in their reports.

This article synthesizes arsenic research along the north-ern half of the Appalachian Mountain belt stretching

Fig. 1. Map of high arsenic belt (left) in the northern Appalachian Mountain belt, Uanthropogenic arsenic sources. The high arsenic in groundwater zone is defined as inthe bedrock geology of the Northern Appalachian Mountain belt, with Silurian angranites highlighted. Terrane names include the Grenville (GRE), Bonson Hill Arc (Boffshore based on Maguire et al. (1999, 2004).

approximately from eastern Pennsylvania northward throughnortheastern Maine, in the United States (Fig. 1). Datareported by individual authors are collectively analyzed inan attempt to connect processes occurring across large spatialexpanses. A geologic model for the origin of arsenic in thebedrock is presented, and the importance of nitrate incontrolling arsenic mobility is suggested. While the primaryfocus of this paper is on naturally occurring arsenic ingroundwater, two examples of arsenic contamination ofsurface waters are also presented. The article concludes bydiscussing generalized processes and mechanisms thatexplain the naturally occurring arsenic concentrations in thegroundwater of the Northern Appalachian Mountain belt,both from a geologic perspective over the last billion years,and from the geochemical perspective of the modern

SA (inset). Specific locations discussed in the text include both geogenic andcluding all of the known geogenic arsenic occurrences. On the right is a map ofd Devonian metasedimentary rocks, Triassic fifth basin rocks, and paleozoicHA, dashed line) and the Central Maine Terrane (CMT). The CMT is extended

10 S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

environment. These processes can then be used as a frame-work for future studies in this region and as a potential modelfor arsenic behavior in similar tectonic and geochemicalenvironments.

2. Arsenic in natural and affected waters

The general processes that influence the behavior ofarsenic in groundwaters around the world have beenreviewed in several excellent publications (e.g. Welch et al.,2000; Smedley and Kinniburgh, 2002). The earliest reports onarsenic in groundwaters within the Northern AppalachianMountain Belt were limited in scope due to the historicpaucity of data on the concentration of arsenic in naturalwaters and geologic materials (Hitchcock, 1878; Myers andStewart, 1956). Local arsenic problems were discovered whena homeowner requested a special test, or hazardous waste sitewas studied carefully (e.g. US-EPA, 1981; Boudette et al., 1985;Zuena and Keane, 1985). Recent contributions have filled thisdata gap by characterizing the regional extent to which arsenicoccurs naturally, complemented by a series of more local studiesthat seek to understand processes occurring within specificproblem areas. Examination of regional trends in arsenicconcentrations has generated two primary results. The first isan accurate estimate of howmany people are exposed to a givenconcentration of arsenic through drinkingwater, and the second,an indication as to whether the arsenic is geologically derived ordue to contamination by anthropogenic activities.

2.1. Regional population exposure estimates

Sampling of water wells throughout the AppalachianMountain Belt combined with water use and populationdistributions provides a reasonable estimate of drinkingwater exposure. In a regional study of arsenic in the centerof the northern Appalachian Mountain Belt, nearly 1000water samples from private homeowner wells had arsenicconcentrations ranged from 0.0003 µg/L up to 180 µg/L(Peters et al., 1999). A parallel study conducted over the widergeographic area of the New England Coastal Basins (NECB) aspart of the national water quality assessment (NAWQA)sampled 58 private bedrock wells and tabulated data from adatabase of over 1000 municipal well tests. Concentrationsacross this wider area ranged from b5 µg/L to approximately55 µg/L (Ayotte et al., 1999, 2003). The percentage of wellswith greater than 10 µg/L arsenic ranged from 17% of theprivate bedrock wells across the New England Coastal Basinsto approximately ~6.3% of all wells in the state of NewHampshire (Ayotte et al., 1999, 2003; Peters et al., 2006). Inboth datasets, arsenic concentrations measured in privatelyowned bedrock wells had median concentrations of arsenicthat were greater than wells screened in surficial depositssuch as sands and gravels (Peters et al., 1999; Ayotte et al.,2003; Peters and Blum, 2003; Peters et al., 2006).

Results from the Pennsylvania/New Jersey area haveshown that the high arsenic area also extends further to thesouth within the Appalachian mountain belt (Burkert, 2005;Serfes, 2005; Serfes et al., 2005; Peters and Burkert, 2007).Water samples collected from 94 households within theNewark basin near the border with Pennsylvania (Fig. 1)showed that 15% of the water sources had arsenic concentra-

tions greater than 10 µg/L, with the maximum concentrationmeasured at 215 µg/L (Serfes, 2005). In adjacent Pennsylvania,water samples from 54 private homeowner wells had amedian arsenic concentration of 2.9 µg/L, and a maximum of65 µg/L with 22% of samples above 10 µg/L (Peters andBurkert, 2007).

The occurrence of arsenic in drinking water throughoutthe Northern Appalachian Mountains provides an opportu-nity to study the effects of chronic low-level ingestion onhuman health (Karagas et al., 1996; Karagas et al., 1998;Karagas et al., 2001; Colt et al., 2002; Andrew et al., 2003;Karagas et al., 2004; Ayotte et al., 2006; Andrew et al., 2006).Increased occurrence of bladder and skin cancer wasobserved at arsenic concentrations greater than 330 µg/L,and in some cases, synergistic with smoking (Karagas et al.,2004).

2.2. Bedrock geologic source of arsenic

Throughout both the New Hampshire arsenic study andthe NECB NAWQA study, arsenic was detected in ground-waters parallel to the Appalachian Mountain belt, typicallyeast of the modern topographic high (Peters et al., 1999;Ayotte et al., 2003; Peters and Blum, 2003; Peters et al., 2006).The outline of a “high arsenic belt” can be constructed andplotted (Fig. 1) from these regional data inclusive of evidenceof natural contamination studied locally. Further to the south,the arsenic belt is not well defined, though it initially appearsto at least include the Newark Basin (Serfes, 2005; Serfes et al.,2005; Peters and Burkert, 2007). This high arsenic belt islocated primarily in crystalline bedrock, with the high arsenicaquifers in this area mostly hosted in fractured rock. Incontrast, many of the high arsenic areas around the world arehosted in young porous media aquifers that routinelyexperience reducing conditions during burial and diagenesis.These conditions can liberate arsenic from iron oxides, and istypically accompanied by high iron concentrations along withlow Eh, no dissolved Oxygen, and low sulfate (Smedley andKinniburgh, 2002). Fractured crystalline bedrock aquiferstypically have no organic matter, and the dominant sourceof regionally reducing conditions occurs via sulfide oxidationand/or nitrate oxidation. Of course local inputs of high nitrateconcentrations from fertilizer or sulfide mineralizations cancreate strong reducing conditions in small areas, but most ofthe high arsenic is observed along with high pH, low iron, andeither high Eh or some measurable amount of dissolvedoxygen. (e.g. Ayotte et al., 2003; Peters and Blum, 2003).

2.3. Local variability in groundwater arsenic concentrations

The earliest reports of arsenic in localized regions did notresult in specific discoveries of source mechanisms, thoughthey did identify locations in the Northern Appalachians withhigh arsenic. Residents of the town of Hudson, NH (Fig. 1)discovered high arsenic in drinkingwater wells in 1981 and asa result, 100 infants were tested for arsenic poisoning (NewYork Times, 1981a,b). Followup studies by the EPA and theUSGS agreed that arsenic was present in the groundwater ofthe area (US-EPA, 1981; Boudette et al., 1985). Residentialwells near the town of Pepperell, MA, were tested due tocitizen concern, and were reported to have 31% of wells

11S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

greater than 20 µg/L and 12% of wells in excess of 50 µg/L(Zuena and Keane,1985). The southernMaine communities ofBuxton and Hollis also had reports of high arsenic, promptinga study of more than 1200 drinking water samples (Marvin-ney et al., 1994). Approximately 29% of the wells had drinkingwater arsenic concentrations in excess of 20 µg/L, and 14% hadgreater than 50 µg/L (Marvinney et al., 1994). Arsenicconcentrations were generally highest in wells located inthe metasedimentary lithologies surrounding granitic intru-sions, and in water from drilled bedrock wells compared todug wells or springs (Marvinney et al., 1994). Further to thesouth, the Connecticut counties of Woodstock, East Hampton,and Colchester were studied by the Agency for ToxicSubstances and Disease Registry (ATSDR) due to reports ofhigh arsenic in drinkingwater. From a set of 50wells sampled,10 contained arsenic in excess of US EPA's MCL of 10 µg/L(ATSDR, 2003).

In some areas, the percentage of wells above the currentUS EPA MCL can be staggeringly high, and in small neighbor-hoods (~1 km2) can approach 100%. Within a watershed inNorthport Maine, 69% of wells exceeded 10 µg/L (Lipfert et al.,2006), and in the Goose River watershed, one third of samplescollected exceeded 10 µg/L (Sidle et al., 2001). Wells adjacentto a sulfide mineralization zone in the Newark basin, andwells in coastal Maine had concentrations as high as 220 µg/L,and 2.0 mg/L respectively. (Serfes, 2005; Lipfert et al., 2006).

3. Sources of arsenic

Examination of anthropogenic inputs, rocks, minerals, andother potential sources of arsenic to groundwater systems iscritical to an accurate understanding the origins of arsenic ingroundwater at both the regional and local scale.

3.1. Anthropogenic sources

For the majority of the late 19th and early 20th century,arsenical pesticides were used in agriculture. The proximity ofthe wells to historic fruit orchards led to the suggestion thatthe arsenic in groundwater was sourced in pesticides (Zuenaand Keane, 1985). Arsenical pesticide application rates arehighest in the Connecticut river valley, coastal Massachusetts,and northernMaine (Robinson and Ayuso, 2004). Comparisonof pesticide use to a map of measured arsenic concentrationsin groundwater along with lead isotope data led the authorsto conclude that geologic materials are the most likely sourceof arsenic to groundwater (Robinson and Ayuso, 2004).

3.2. Geologic sources

3.2.1. RocksThe belt of high arsenic concentrations in groundwaters is

centered within rocks of the Central Maine Terrane (CMT), anaccretionary block representing marine basin fill sediments(Ayotte et al., 1999; Peters et al., 1999; Ayotte et al., 2003;Peters and Blum, 2003; Peters et al., 2006). The highestgroundwater concentrations within the CMT are locatedadjacent to geologic contacts between the metasedimentaryrocks and intrusive bodies (Peters et al., 1999; Lipfert et al.,2006). Analysis of Silurian and Devonian rocks from theseareas showed that they contained 1 M HNO3 acid leachable

form of arsenic at approximately 60 mg/kg, the mineral formsof which will be discussed in the subsequent section onmineralogy in this paper (Peters et al., 1999). Near the town ofNorthport, concentrations are highest (68 mg/kg) within thePenobscot formation, a sulfidic carbonaceous, thinly beddedalternating schist and quartzite with rare limestone andcalcareous sandstone (Lipfert et al., 2006).

Nearby, the source rocks for the high arsenic measured inthe Goose River Basin are the Bucksport formation, a fine-grained 5-cm-bedded to massive, medium brownish grayquartz-plagioclase-biotite-hornblende granofels, with 2 to10 cm beds of medium greenish gray calc-silicate granofels(Sidle et al., 2001). In the same area, the anatectic Waldoborogranite (~368 Myr), a garnet bearing two-mica foliatedgranite to granodiorite, is likely to be derived from this unit(Sidle et al., 2001; Tucker et al., 2001; Hussey and Berry, 2002;West et al., 2008). The granite has higher median arsenicconcentrations (46 mg/kg) than the surrounding Bucksportformation (39 mg/kg) (Sidle et al., 2001). Measured ∂18OSO4

values in the waters containing high arsenic are distinct andmost closely match those values measured in the crystallinerocks, and not that of the overburden (Sidle, 2002).

3.2.2. MineralsThe primary minerals commonly credited with originally

sourcing high concentrations of arsenic to groundwatersinclude arsenopyrite (FeAsS) and arsenian pyrite (Peters andBlum, 2003; Barnard, 2006; Lipfert et al., 2006; Peters et al.,2006). However, there are hundreds of other secondaryarsenic minerals that may be found in rocks, particularly inhydrothermal alteration zones, and the reader is referred toSmedley and Kinniburgh (2002) for a more complete listing.

Arsenopyritewas described as the primary arsenicmineralretrieved in drill cores within the area containing the highestconcentrations of arsenic in groundwater from Northport,Maine (Lipfert et al., 2006). In some locations, lenses ofarsenopyrite grains were found to reach thicknesses of 3 cm(Lipfert et al., 2006). Generally speaking, other sulfideminerals in arsenic rich rocks from Northport Maine arequite low in total arsenic (Lipfert et al., 2006). Arsenic richpyrites are sometimes described (Lipfert et al., 2006), buttheir overall total arsenic concentration is usually less than 5%w/w, and in many cases between 0.1 and 2.0% w/w (Sidleet al., 2001; Lipfert et al., 2006). Loellingite (FeAs2) andCobaltite (CoAsS) are observed in some rocks from the southcentral Maine coast (Barnard, 2006).

3.2.3. Nanocrystalline phasesClose examination of the mineralogy of an elevated

groundwater arsenic region in central New Hampshireindicated that the mineralogic forms previously characterizedas macroscale minerals were actually assemblages of nanos-cale mineralogic forms. Using transmission electron micro-scopy (TEM) and high-angle annular dark field scanningtransmission electron microscopy (HAADF-STEM), a zone ofvarying chemistry was discovered along some arsenopyritegrain boundaries (Utsunomiya et al., 2003). These grainboundaries exhibited surface parallel zonation of variousmetals, including U and up to 15% w/w Cu, indicative ofalteration by three different hydrothermal fluids (Utsunomiyaet al., 2003). Typical elements observed in these altered

12 S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

arsenopyrite grains included As, Fe, K and O, but with variableFe/As ratios ranging from 1.15 to 1.6. Closer examination ofthis altered mineral revealed an assemblage of 10–50 nmnanocrystals and amorphous phases instead of a singlecrystalline phase (Utsunomiya et al., 2003). The measuredd-spacing and stoichiometry for these minerals correspondedto magnetite (Fe3O4) and westerveldite (FeAs). Westervelditeis commonly found in associationwith chromite–niccolite orein schieren and in ultrabasic rocks, and often with highconcentrations of nickel and cobalt substituting for iron (Oenet al., 1972). The discovery of nanocrystalline forms and themineral westerveldite was unexpected. The hypothesizedparagenesis reactions were proposed as 1) initial arsenopyriteformation, 2) hydrothermal alteration with potassiumenriched fluids converting arsenopyrite to westerveldite andmagnetite with loss of sulfur, and 3) low temperatureoxidation and weathering decomposing westerveldite andreleasing arsenic to solution (Utsunomiya et al., 2003). Theseobservations highlight the complexities of arsenic miner-alogies that impact groundwater.

3.2.4. Soils and stream sedimentsSoils and stream sediments can act as both sources and sinks

for arsenic. In the Goose River basin, the glacial overburden hadarsenic concentrations exceeding20mg/kg in10%of the samples,with one high sample having approximately 35 mg/kg arsenic(Sidle et al., 2001). At the Coakley landfill site in NewHampshire,a glaciomarine clay with approximately 20 mg/kg arsenic isthought to be the source of the aqueous phase arsenic ingroundwater (deLemos et al., 2006). In this clay layer, the arsenicis present as arsenic(V) bound to ferrihydrite as determined byEXAFS and XANES (deLemos et al., 2006). In a comprehensiveexamination of stream channel sediments across all of NewEngland, Robinson andAyuso (2004) found that 13% of all streamsamples are above 10 mg/kg, the threshold effect concentrationfor benthos. Lead isotope measurements on sediment leacheswhere arsenic and lead were correlated showed that most of the

Fig. 2. Plot of arsenic concentration as a function of pH in localized high arsenic concefrom Lipfert et al. (2006). The Central NH data are 32 wells from Peters and Blum (between undifferentiatedmetasedimentary units (Mu) and calcareousmetasedimenof 0.02 μmol/L) are plotted. Dashed line is the pH dependent leaching of core sample2000). Regressions through each dataset all have positive trends though the degree

arsenic is likely derived from natural bedrock sources, and notfrom lead arsenate pesticides (Robinson and Ayuso, 2004). In afew of the most dominant agricultural areas, slightly radiogeniclead isotope compositions indicate possible contributions fromlead arsenate pesticides (Robinson and Ayuso, 2004).

4. Mobilization mechanisms

In many cases, the heterogeneous nature of groundwaterarsenic concentrations is puzzling. Wells located less than100m apart, with screened intervals in the same geologic unitsat the same depth, can have dramatically different arsenicconcentrations (Boudette et al., 1985; Zuena and Keane, 1985;Marvinney et al., 1994; Peters and Blum, 2003; Lipfert et al.,2006). The cause of this heterogeneity is still not completelyunderstood, though hypotheses about groundwater mixing,reaction zones around wells, water from different fracturesystems, and preferredflowdirections in fractured rock have allbeen proposed. Multiple mechanisms that control the dissolu-tion and/or desorption of arsenic from the solid phasemay helpexplain the observed heterogeneous distribution. A challengingaspect to evaluating these mechanisms is water that may havenever contacted arsenic sourcematerials will have awide arrayof values hypothetically linked to the suspected controllingmechanism. For example, if pH dependent desorption is asuspected mechanism, we would predict that as pH increases,arsenic concentration also increases. However, not all high pHsamples will have high arsenic concentrations. The high pHvalues with low arsenic may be the result of never contactingarsenic bearing materials. There is no straightforward solutionto this problem other than to closely examine the raw data andnot rely solely on t-tests and ANOVA tables.

4.1. pH dependent desorption

The most common relationship observed among theregional datasets is that water samples with high arsenic

ntration groundwater areas. The coastal ME data are the 10 high-arsenic wells2003). The NECB data are from Ayotte et al. (2003) retaining the distinctiontary rocks (Mc). Within these units only data that are above the detection limis containing arsenic that were obtained from southeast Michigan (Kim et al.of fit is variable.

t,

13S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

typically also have pH values greater than approximately 6.5(Ayotte et al., 2003; Peters and Blum, 2003). The most oftencited reason for this trend is pH control of arsenic adsorptionto mineral surfaces, particularly arsenic (V) to iron minerals(Peters et al.,1999; Ayotte et al., 2003; Peters and Blum, 2003).In both New Hampshire and the NECB data, pH values rangedfrom 5 to 9, with a rise in dissolved arsenic concentrationsobserved at about pH 6.5 (Ayotte et al., 2003; Peters and Blum,2003). Arsenic concentration data for New Hampshire, NECB,coastal Maine, and New Jersey are plotted vs pH in Fig. 2. Thegeneral trend within each dataset is to have lowest pH values

Fig. 3. Plot of arsenic concentrations as a function of iron concentrations in localizedbeen described to come from arsenic impacted landfill sites. In most naturally occurrboxed area in the lower left corner of the upper plot (A) is replotted with an expan

containing the lowest arsenic concentrations. At higher pHvalues, the arsenic concentration can be either high or low,most likely due to variations in source rocks. The parallelslope of most of the best-fit lines through the data indicates asimilar pH dependent adsorptive process. The Newark Basindata do not follow the same trend in slope for unknownreasons. The vertical offset between each of the lines could bedue to multiple factors, two of which include: the character-istics of the adsorbing mineral surface and the composition ofcompeting counter-anions in solution, for example Fe(II)(Appelo et al., 2002). For comparison, a dashed line is plotted

high groundwater arsenic areas. Waters with As:Fe ratios less than 1:100 haveing arsenic areas, As:Fe ratios are much higher, in some cases with AsNFe. Theded scale in the lower plot (B).

14 S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

that represents rock core samples that containing arsenopyr-ite that were leached with synthetic groundwater with0.04 M of total inorganic carbon species at various pH values(Kim et al., 2000). The authors hypothesized that carbonatecomplexes are assisting in the dissolution of the rockmaterials, but it is also possible that the carbonate leachsolutionmaintained the highest pH, preventing re-adsorptionto the core materials following initial dissolution. The similarslope between the laboratory leaching experiment and foursets of field data strongly suggests that in areas with arsenicavailable in the bedrock geology, pH exerts the strongestcontrol on dissolved arsenic concentration.

A plot of the iron and arsenic concentrations is often usedto discern any relationship between these two elements(Fig. 3). In some cases, the highest concentrations of arsenicare matched by equimolar amounts of iron, indicated by a 1:1line near the y-axis of Fig. 3A. Samples with this signaturehave several potential explanations. First is the stoichiometricdissolution of arsenopyrite followed by conservative trans-port of both the iron and arsenic. The second is pH dependentdesorption of arsenic(V) in the presence of some dissolvediron. In Fig. 3A, the samples falling along the 1:1 line (filledcircles) are from Coastal Maine (Lipfert et al., 2006) and areconsistent with pH dependent desorption (Fig. 2). Also fallingalong the 1:1 line are some samples from Central NewHampshire (open circles, Peters and Blum (2003)), and somesamples from the NECB (open triangles, Ayotte et al. (2003)).Not all samples plot along the 1:1 line, indicating some of thehigh arsenic in those samples may be due to reductivedissolution.

4.2. Reductive dissolution of arsenic bearing oxides

Arsenic adsorbed to and/or incorporated within ironoxides can be liberated during reduction of the iron phasesresults in liberating the arsenic to solution. The ratio ofarsenic:iron in precipitated iron oxide minerals is likely to bequite low, perhaps ranging only as high as 1:10 (Pichler et al.,1999; Smedley and Kinniburgh, 2002). Dissolution of theseiron oxides will likely result in iron concentration in excess ofarsenic concentration, therefore plotting well below the 1:1line on Fig. 3. Some sites exhibit clear mechanisms ofreductive dissolution, typically with iron concentrations inexcess of 200 µmol/L (11 mg/L) and negative Eh values (e.g.deLemos et al., 2006). These sites plot along the X axis ofFig. 3, with ratios of arsenic:iron less than 1:100 (×'s, Fig. 3B,deLemos et al. (2006)). High arsenic samples that plot in themiddle of Fig. 3 are likely to be released via reductivedissolution of iron oxides as well as perhaps some pH drivendesorption.

4.2.1. Anthropogenically enhanced releaseAnthropogenic activities that change the redox state of

groundwater can influence the stability of arsenic bearingminerals and therefore the arsenic concentrations in ground-water. The emplacement of landfills with redox-active wastematerials is one way that human activities can impactgroundwater redox. In these cases, the potentially responsibleparty is eager to ascertain whether the landfill is responsiblefor the contamination. Unfortunately there is not always aclear answer to this question. For example, leachate may

contain high arsenic, and might be clearly traceable to thelandfill as the source. However, in other cases, the redoxchanges or pH changes induced by leachate movement maycause arsenic mobilization from a natural geogenic source. Inat least three locations of the northern Appalachian Mountainbelt, landfill affected redox conditions are thought to initiatethe release of arsenic from geologic materials (Stollenwerkand Colman, 2002; Stollenwerk and Colman, 2004; Keimo-witz et al., 2005a,b; Mayo, 2006; deLemos et al., 2006).

In the seacoast region of New Hampshire, arsenic wasmeasured in proximity to the Coakley landfill, a site currentlyon the United States Comprehensive EnvironmentalResponse, Compensation, and Liability Act (CERCLA) NationalPriorities List (NPL) list (deLemos et al., 2006). Site remedia-tion included capping of the waste pile to eliminate waterflow through the pile and continuous monitoring for naturalattenuation. After the capping was complete, benzeneconcentrations decreased over time, from concentrations ofapproximately 10 µg/L down to 4 µg/L, a 65% decrease(deLemos et al., 2006). Over the same time period, arsenicconcentrations increased from 14 µg/L up to 22 µg/L, a 58%increase (deLemos et al., 2006). The change in benzeneconcentrations is negatively correlated with arsenic concen-trations (R2=0.86) suggesting the two processes could belinked (deLemos et al., 2006). In groundwater at the site, ironand arsenic concentrations are generally correlated, thoughthe maximum iron concentration is controlled by sideritesolubility (deLemos et al., 2006). The source of the increasingarsenic concentrations was attributed to iron reduction andsubsequent arsenic release from a glaciomarine clay layer.Incubation experiments indicate that the iron reduction islikely to be bacterially mediated and stimulated by theaddition of glucose (deLemos et al., 2006).

In South Central Maine, a similar landfill was closed afterVOCs were detected in a residential well off-site (Lackovicet al., 1999; Nikolaidis et al., 2004; Keimowitz et al., 2005b,c).After inclusion in the CERCLA NPL in 1983, the pile received alow permeability cap and both vapor and groundwaterextraction systems were installed to lower the VOC concen-trations in the subsurface (Lackovic et al., 1999). As VOCconcentrations declined (~10 000 µg/L to ~10 µg/L), arsenicconcentrations slowly increased (~200 µg/L to ~400 µg/L)(Keimowitz et al., 2005b). Other solutes typical of landfillleachate do not correlate with the increased arsenic and themeasurement of elevated concentrations upgradient indicatethat the arsenic is not sourced in the landfill materials(Keimowitz et al., 2005b). The most likely source is theunconsolidated glacial materials, including tills and outwash.Concomitant increases in both dissolved iron and arsenictowards the center of the affected area suggest that reductionof iron oxides is releasing the arsenic to groundwater(Keimowitz et al., 2005b). This was confirmed with anoxicincubations of surficial materials that released arsenic into thewater at a rate of 1.7 mg arsenic per kilogram of sediment(Keimowitz et al., 2005b).

A number of landfill sites in central Massachusetts areassociated with arsenic concentrations in excess of 5000 µg/L.At one site, arsenic concentrations in groundwater increasefrom less than 1 µg/L upgradient to greater than 4000 µg/Ldowngradient of the waste pile (Mayo et al., 2003; Mayo,2006). Increases in arsenic concentration correlate well with

Table 1Statistical results of comparing arsenic concentrations in drinking water withvarious potential indicator species

pH Nitrate Chloride Na/Ca Mechanism

NECB, Mc 0.05 0.003 0.56 0.3 pH dependentadsorption

NECB, Mu 0.01 0.003 0.3 0.8 pH dependentadsorption

Central NH 0.0002 0.05 0.0001 0.6 pH dependentadsorption

Coastal ME, High 0.25 0.23 0.3 0.03 Exchange, flowpathCoastal ME, All 0.08 0.0001 0.002 0.08 Exchange, flowpathNewark Basin 0.0001 0.0003 0.003 0.04 VariousAggregate 0.0001 0.0001 0.0001 0.025

All p-values are computed from Spearman's Rho rank correlations which areapplicable for non-parametric data. The NECB and Central NH data showevidence of pH dependent control on adsorption to mineral surfaces. TheCoastal ME data show a stronger relationship to Na/Ca, which is partiallyindicative of degree of interaction with bedrock. Data from the Newark basinshow both pH dependent adsorptionmechanisms along with some degree offlowpath exchange.

15S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

decreases in ORP and increases in dissolved iron concentra-tion (Mayo, 2006). Increases in arsenic do not correlate wellwith chloride or sodium, suggesting a source other thanlandfill leachate (Mayo, 2006). The authors conclude that thepresence of reducing conditions in the subsurface is driven bylandfill leachate, which then destabilizes iron oxides contain-ing arsenic (Mayo, 2006).

In all of these cases, the local redox environment isstrongly affected by landfill leachate. In some cases this isthought to be the increased availability of a metabolicallyavailable carbon source fueling microbial activity, or acomplex array of processes that create a locally reducingenvironment. In all cases, high arsenic is accompanied by veryhigh iron concentrations, and likely represents reductivelydissolved iron oxides that contained arsenic.

4.2.2. Bacterially mediated reductionArsenic reducing bacteria have been isolated from regions

in the northern Appalachian Mountain belt. The bacteriumwas cultured from a water sample obtained from a ground-water well located in Northport, Maine that was known tohave an arsenic concentration of ~1400 µg/L (Weldon andMacRae, 2006; MacRae et al., 2007). The 16 S rRNA from thecultured organism was amplified, sequenced, and analyzedusing BLASTn, and submitted to Genbank. The isolate had noidentical matches in the database, but was closely related toseveral Sulfurospirillum species. Growth experiments showedthat this bacterium could utilize arsenic as a terminal electronacceptor thereby reducing the arsenic from As(V) to As(III).The organism requires, and is likely limited by, an organiccarbon substrate for growth (MacRae et al., 2007).

4.3. Oxidation of sulfides

Arsenic bearing sulfide minerals are the most commonlycited primary source for high arsenic groundwaters in theNorthern Appalachian mountain belt (Peters et al., 1999; Sidleet al., 2001; Sidle, 2002; Serfes et al., 2005; Lipfert et al.,2006). As such, most studies attribute initial release of arsenicto oxidation of these sulfides followed by dissolution. Duringthe oxidation of pyrite in the Goose River basin, the resulting∂18OSO4 is indicative of the primary oxidant being atmo-spheric dissolved oxygen (Sidle, 2002). Some of theseminerals can occur as nanocrystalline phases, which makesprediction of their behavior challenging. Due to their highinherent surface energy, nanocrystalline phases can be lessstable than thermodynamic predictions based on larger sizeparticles, and can dissolve under what appear to be stableconditions (Stumm and Morgan, 1996; Utsunomiya et al.,2003).

4.4. Reduction of nitrate

The presence of nitrate in groundwater may play a role inoxidation of arsenic, similar to what has been observed forsurface water (Senn and Hemond, 2002). Within each regionaldataset, nitrate and arsenic were statistically correlated(Spearman's Rho), with p-values generally less than 0.05(Table 1). Higher concentrations of nitrate correspond tolower concentrations in all datasets (Fig. 4). Nitrate has beenshown to act as a terminal electron acceptor in surface water

systems and promotes the formation of arsenic(V) and Fe(III)even under anoxic conditions (Senn and Hemond, 2002). Ingroundwater systems, dissolved oxygen can be consumed byrespiration and since it is an environment that is isolated fromthe atmosphere, oxygen is not easily replenished. A field scaletest of arsenic mobility controlled by nitrate in Cape Cod,Massachusetts, illustrated the process that nitrate caused theformation of iron oxides and adsorption of arsenic to those ironoxides (Hohn et al., 2006). When the source of nitrate wasremoved, the groundwater redox environment returned to pre-experimental conditions with iron(II) and arsenic(III) as thedominant species (Hohn et al., 2006). Nitrate reductionprovides amechanism that explains the presence of iron oxidesin anenvironmentwhere theymight not otherwise be stable. Insome arsenic rich environments, arsenic behavior is generallyconsistent with adsorption to iron oxides, though seeminglyunder partially anoxic conditions (e.g. Lipfert et al., 2006).

4.5. Groundwater chemical evolution

Evolution of water along flowpaths has been invoked toexplain the distribution of arsenic concentrations (Sidle,2003; Sidle and Fischer, 2003; Lipfert et al., 2006; Peterset al., 2006). Chloride is slightly negatively correlated withhigh arsenic in non-coastal areas (Table 1). There are severalinterpretations for the presence of chloride in these ground-waters, including contributions from shallow flowpaths thatare in contact with road de-icing salts and remnant salts frommarine inundation after the Wisconsinan glaciation (Peterset al., 1999; Lipfert et al., 2006). However, none of these highchloride explanations justifies why arsenic should be nega-tively correlated with chloride. Positive correlations withchloride might indicate evaporative concentration, such asobserved in arid regions, though even in those samples, thetrends are complex (Welch and Lico, 1998).

Directmeasurement of groundwater age using a combinationof CFC, 3H, and 85Kr methods show that 90%+ of high arsenicgroundwaterwells inMaine andNewHampshire have post 1950groundwater ages (Ayotte et al., 2003; Sidle and Fischer, 2003).Neither study showed arsenic concentrations to be a directfunction of age, but rather that nearly all high arsenic waters are

Fig. 4. Plot of arsenic concentrations as a function of nitrate concentrations in high groundwater arsenic areas. A significant negative correlation is observed(pb0.0001). p-values are listed in Table 1.

16 S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

moderately young (b50 years old) (Ayotte et al., 2003; Sidle andFischer, 2003). Groundwater contact time and solution maturityhas been suggested as a possible mechanism for explaining higharsenic in some areas (Lipfert et al., 2006). Exchange of calciumfor sodium has been suggested as a reasonable proxy forgroundwater maturity, and correlates with increasing arsenicconcentration (Lipfert et al., 2006). Aggregating the data for theentire Appalachian mountain belt, the trend becomes less clear,but remains significant for thewater samples from coastalMaineand also for the samples from the Newark basin (Table 1).

4.5.1. Well depthArsenic concentrations do not appear to relate to well

depth within a particular geologic unit (Sidle and Fischer,2003), however, dramatically different concentrations areobserved between shallow wells in unconsolidated depositsand those tapping deeper fractured crystalline aquifers (Peterset al., 1999; Peters and Blum, 2003). The complex fracturegeometry andmultiple flow directionsmakes simple relation-ships with parameters such as well depth implausible andunlikely.

5. Discussion

5.1. Arsenic mobility during tectonic assembly of Eastern NorthAmerica

Arsenic in rocks is found across multiple terranes,including the Central Maine Terrane in New Hampshire(Peters et al., 1999; Ayotte et al., 2003), and the NewarkBasin (Serfes, 2005; Peters and Burkert, 2007). The geologicconstruction of the eastern margin of North America isimmensely complex, and as such, the focus of considerableongoing study. In the following section, the generalizedsequence of events is outlined based on summaries and

references included in (Hatch et al., 1984; Keppie, 1993; Faill,1997a,b), highlighting events and processes that may haveinfluenced the movement of arsenic in the NorthernAppalachian mountain belt.

The earliest well constrained collisional event initiated inthe Mesoproterozoic with the Grenville Orogeny thatassembled the supercontinent Rodinia (Fig. 5A). Subsequentrifting generated extensional half-graben blocks and even-tually opened the Iapetus Ocean, which persisted from theNeoproterozoic through to the Cambrian (Fig. 5B). During thistime, the eastern margin of Laurentia (now North America)was passive, with a volcanic island arc present some distanceout in the Iapetus Ocean (Fig. 5C). During the OrdovicianTaconic orogen, several terranes were accreted to the easternmargin of Laurentia, including the Brandywine Microconti-nent in the central Appalachians and the Hawley andAmmonoosuc/Bronson Hill Arc (BHA) in New England(Fig. 5D). Subsequent erosion off the high mountain rangesformed during the Taconic orogeny were deposited along theflanks of the mountain belt into the Theic ocean during theSilurian, Ordovician, and Devonian. The composite micro-continent of Avalonia was situated to the east of theLaurentian coastal margin, and was likely depositing sedi-ment to the east and west (Fig. 5D). Activation of a doublesubduction zone beneath both Avalonia and Laurentia closedthe Theic ocean during the Devonian Acadian orogeny(Fig. 5E). Sediments deposited between Avalonia and theBHA were metamorphosed and emplaced onto Laurentia asthe rocks of the CMT. Crustal thickening and heat productioninduced partial melting at the core of the mountain belt,initiated hydrothermal systems and produced anatectic meltsof accreted crustal material (Fig. 5F). Subsequent collision ofGondwana during the Alleghenian orogeny accreted morecrust onto Laurentia to form Pangea (Fig. 5G). Rifting ofPangea opened the Atlantic ocean in the Triassic (Fig. 5H),

Fig. 5. Diagrammatic evolution of high arsenic belt in Northern Appalachian Mountains. Drawings A–G represent the evolution of the Northern Appalachians.Figures H–I depict possible relationships that could explain the high arsenic in the transition zone between Northern and Southern Appalachians. Abbreviations arekeyed to Fig. 1. Cross section diagrams after Marshak (2001).

17S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

creating extensional fault bounded half grabens on NorthAmerica, including the Newark Basin (Fig. 5I).

The primary location of the high arsenic belt (Fig. 1) inNew England is within the CMT. While we cannot definitely

prove the precise origin of the arsenic rich sediments withinthe CMT, we can provide one plausible explanation. The CMTis predominantly composed of Siluro-Devonian sedimentsoriginally deposited in the submarine environment between

18 S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

Avalonia and the BHA, with mixed provenance from the eastand west. From the west, the source of arsenic rich sedimentscould possibly have included the Grenville Basement ofLaurentia and also the newly accreted BHA. If arsenic bearingsediments had been sourced from either the margin ofLaurentia, or off the BHA, then high groundwater arsenicwould be found in sediments to the west of the BHA. IfLaurentia is the source of the high arsenic sediments, thenwewould also expect to find a well defined trend of high arsenicalong the Appalachian Mountain belt to the south. Currentdata suggests groundwater arsenic is low to the west of theBHA and also low in the southern Appalachian Mountains(Peters et al., 1999; Ayotte et al., 2003). Therefore the source ofthe arsenic enriched sediments is most likely to have been tothe east, probably from Avalonia.

Avalonia may have contained arsenic enriched materialsthrough a variety of mechanisms, and here, we present onesuch scenario. During the time before accretion, Avalonia wasadjacent to a subduction zone, and it is possible that thenearshore environment was hydrothermally active and mayhave been depositing arsenic via submarine vents. Thedelivery of arsenic enriched hydrothermal fluids could haveoccurred in a similar fashion to those described in themodernnearshore marine environments of the oblique continent–arccollision in Papua New Guinea (Pichler and Veizer, 1999; Priceand Pichler, 2005). Hydrothermal fluids are delivering 1.5 kgper day of arsenic to the nearshore environment as arsenic(III)which rapidly oxidizes to arsenic(V) and is adsorbed on ironoxides and other oxic mineral phases (Price and Pichler,2005). Measured arsenic concentrations in nearshore sedi-ments are greater than 500 mg/kg (Price and Pichler, 2005).While themineralogic composition of the original arsenic richsediments on Avalonia is not known, the highest concentra-tions of arsenic are found in association with calc-silicatemetapelites (Ayotte et al., 2003). The connection to calc-silicates has perplexed arsenic researchers due to the lack of amechanistic connection to any calc-silicate mineral orprocess. However, the connection to calcareous sedimentsmay simply be from the coincident deposition of high arsenicsediments proximal to shallow nearshore calcareous reefsystems (Pichler et al., 1999). An alternate explanation to thehydrothermal deposition of arsenic is the sequestration ofreduced arsenic sulfides in a euxinic marine basin thatperiodically appeared in the Theic Ocean prior to Avaloniancollision. This explanation is not favored due to the lowoccurrence of arsenic in the sulfidic, graphidic schists of NewHampshire and the observed lack of correlation betweendrinking water arsenic and sulfide metasediments in NewHampshire and Maine (Ayotte et al., 1999; Peters et al., 2006).

The widespread high arsenic belt (Fig. 1) contains a fewlocalities where arsenic concentrations in rocks and ground-waters are considerably higher than elsewhere in the belt. Thespecific localities with the highest arsenic concentrations ingroundwaters are usually located along fault boundaries,where hydrothermal fluids can easily travel (e.g. Lipfert et al.,2006) or adjacent to granitic plutons where partial meltingand fractional crystallization would concentrate arsenic intolate stage igneous rocks (e.g. Peters et al., 1999). Whilehydrothermal activity can explain the highly concentratedzones containing extremely high arsenic concentrations, thewidespread presence of high arsenic throughout the CMT

cannot be explained this way. Metamorphic grade varies fromgranulite facies in New Hampshire (Spear et al., 2002) to sub-chlorite grade in northeast Maine (Guidotti et al., 1991), andmetamorphic grade does not correlate with the presence ofarsenic (Ayotte et al., 1999; Peters et al., 1999). Instead, themost widespread regional trends in arsenic concentration canbest be explained by the protolith for the modern CMT.

A summary of the proposed geologic mechanism forarsenic occurrence in New England is the following:

1) Erosion and deposition of arsenic bearing sediments,possibly hydrothermally enriched, west from Avaloniainto the Theic Ocean.

2) Collision of Avalonia and accretion of arsenic enrichedmarine sediments onto Laurentia.

3) Hydrothermal alteration of fault zones and anatectic melt/fractional crystallization of granite pegmatites as enrich-ment mechanisms for the highest arsenic zones.

Further to the south, the presence of arsenic in the NewarkBasin appears confounding at first. The Newark Basin is a faultbounded Tertiary closed rift basin 190 km long and 50 kmwide filled with alluvial fans and fluvial/lacustrine depositsshed from nearby mountains (Faill, 1973; Schlische, 1992;Olsen et al., 1996). Compared to the CMT, the Newark Basin isextensional not collisional, freshwater, not marine, andTriassic, not Devonian. However, we hypothesize that thereis a potential link between these two arsenic enriched areas.The geometries of alluvial fans filling the Newark Basinindicate dominant sediment provenance from the southeast,with lesser northwest materials derived from the over-steepened footwall relay ramps (Schlische, 2003). The precisegeometry of the collisions and accretions to the southeast ofthe Newark Basin occurs in the transition zone between theNorthern and Southern Appalachians and is poorly under-stood. This is partially due to the removal of nearly all of thecollisional materials by rifting or erosion and subsequentcover by the coastal plain sediments of New Jersey (Faill,1997b). However, recent magnetic and gravity studies suggestthe presence of the rocks of the CMT to the southwest of theNewark Basin (Maguire et al., 1999, 2004). We thereforesuggest that it is possible that some of the sediments of theCMT have been reworked and deposited to the Newark Basin.The dissolved arsenic, transported in streams to the closedbasin would then accumulate in the fine grained fluvial/lacustrine deposits of the Lockatong and Passaic formationscurrently observed in the Newark Basin (Serfes, 2005; Petersand Burkert, 2007). During active periods of the CentralAtlantic Magmatic Province (CAMP), periods of hydrothermalfluid activity (Witte and Kent,1991; Kodama et al.,1994) couldexplain the focusing of extremely high concentrationsmeasured along faults, such as those observed near Hopewell,New Jersey (Serfes, 2005).

To the north of the Canada/United States border, there areno reports of widespread naturally occurring arsenic ingroundwater (Wang and Mulligan, 2006). Reported cases ofarsenic in ground and surface waters include geographicallylocal highly mineralized zones, mining and mine tailings, andsoils adjacent wood preservative facilities (Wang and Mulli-gan, 2006). The presence of mineralized arsenic sulfide zonescould indicate that there are other regions of high arsenic due

19S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

to geologic sources, and they simply have not been describedand mapped yet.

The proposed mechanism of arsenic enrichment in theNorthern Appalachian Mountain belt is generally consistentwith shallow crustal recycling of arsenic as an incompatibleelement. Extending these principles to other regions in theworld should include island arcs and arc–continent collisionalenvironments, as well as continent–continent collisions.Several well known examples fit this model, including Taiwan(Tseng et al., 1996; Chang et al., 1999), Japan (Kondo et al.,1999), and perhaps even the rocks of the Himalaya asprotolith for the Bengal fan (Nickson et al., 2000).

6. Surface waters

This review would be incomplete without the mention oftwo significant occurrences of anthropogenic arsenic con-tamination in the shallow groundwater and surface waters ofthe region. First is the Woburn, Massachusetts, Industriplex/Wells G&H site, and second is the Vineland Chemicalcompany site in southern New Jersey. Both sites have actedas natural laboratories where the dynamic interactions ofbiogeochemically active elements such as arsenic, iron, andorganic compounds can be examined.

The city of Woburn, Massachusetts is located approxi-mately 20 km northwest of the city of Boston. The surround-ing Aberjona Watershed has been affected by anthropogeniccontamination of arsenic from industrial sources, in part bythe largest arsenical pesticide manufacturer in the UnitedStates (Davis et al., 1994; Aurilio et al., 1995; Hemond, 1995;Ryan,1979; Knauer et al., 2000; Senn and Hemond, 2002). Theprimary source within the watershed was a waste pile inexcess of 300 metric tons, with nearby soils having arsenicconcentrations up to 30 g/kg (Aurilio et al., 1995; Hemond,1995). Roasting of pyrite to generate sulfuric acid wasattributed to be the primary arsenic waste generatingmechanism, amounting to approximately 170 tons of arsenic.The flow of shallow groundwater through the watershed andinto local streams and retention ponds is slowly moving thearsenic downstream towards the Mystic Lakes (Hemond,1995). At the Halls Brook Storage Area (HBSA), a flood controlpond, springs have been observed to be forming orange-redsediments, most likely iron oxides. Across the length of HBSA,arsenic concentrations decrease from 80 µg/L as mostly As(III)at the upstream end to generally less than 3 µg/L as mostly As(V) at the downstream side (Aurilio et al., 1995; Hemond,1995). From HBSA, the Aberjona river enters Upper MysticLake (UML), where seasonal anoxia and hydraulic complexitycreates heterogeneous geochemical conditions (Knauer et al.,2000; Senn and Hemond, 2002). Interestingly, the oxidationof both iron and arsenic in UMLmay be controlled partially bynitrate (Senn and Hemond, 2002). The oxidation andprecipitation of iron produces particulates conducive toarsenic adsorption, and the oxidation of As(III) to As(V)transforms the arsenic oxyanion into its particle reactive form(Senn and Hemond, 2002).

The city of Vineland is located in southern New Jersey,approximately 65 km southeast of Philadelphia, Pennsylva-nia. The leaching of arsenic biocides from the VinelandChemical Company introduced large-scale arsenic contam-ination to shallow groundwater and surface waters. Con-

centrations of arsenic in surface streams were historicallymeasured at upwards of 10,000 µg/L (Keimowitz et al.,2005b). By 2004, surface water arsenic concentrations haddropped considerably, ranging from 1.7 µg/L upstream to20 µg/L downstream of the site. Sediments at anthropogeni-cally contaminated sites, such as the Vineland ChemicalCompany, can act as host reservoirs, and contain dramaticallyhigh concentrations of arsenic compared to pore waters andsurface waters. In sediments retrieved from the BlackwaterBranch of the Maurice River in New Jersey, arsenic concen-trations in the porewater ranged upwards of 1650 µg/L and inthe solid phase to 1350 mg/kg (Keimowitz et al., 2005b).Depending on the precise balance of redox and pH chemistry,arsenic can be released to solution. Based on carefulexamination of the sediments in the Blackwater Branch,sulfur chemistry appears to control the abundance ofdissolved arsenic as As(III). Approximately 10 km down-stream a dam forms Union Lake, where adsorption to ironoxides controls dissolved arsenic concentrations as As(V)(Keimowitz et al., 2005b).

7. Conclusions

The occurrence of high concentrations of arsenic in thenorthern half of the Appalachian Mountain belt is bestexplained through a crustal recycling model that concen-trates arsenic into shallow crustal materials over the last1 billion years, particularly as sulfide minerals such asarsenopyrite and arsenian pyrite. Oxidation and dissolutionof these primary sulfides release arsenic into generally oxicwaters. The concentration of arsenic oxyanions in the oxicwaters is then modulated by adsorption on iron minerals,which is in turn governed by pH. While this seems to holdtrue for the large regional datasets, a number of local sitestudies demonstrate that reductive dissolution of arsenicbearing iron minerals can also be the final controllingmechanism, particularly related to anthropogenic activities.In all the cases of reported arsenic in groundwater, only twosites are known to have primary anthropogenic arsenicsources. Those two sites are both industrial complexes thathave allowed arsenic wastes on the Earth's surface toinfiltrate to groundwater. Health studies conducted acrossthe region have shown that arsenic in groundwater is not alife threat, and has health consequences only in concentra-tions above 350 µg/L and then only in conjunctionwith otherhealth risks, such as smoking.

Acknowledgements

The author thanks the editors of this special issue, inparticular for inviting this contribution. The author isindebted to the many researchers who have published theirdata along with their papers, without which this reviewwould have been considerably more difficult. Input from twoanonymous reviewers and the editors of this issue helpedimprove the manuscript.

References

Andrew, A.S., et al., 2006. Arsenic exposure is associated with decreased DNArepair in vitro and in individuals exposed to drinking water arsenic.Environmental Health Perspectives 114 (8), 1193–1198.

20 S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

Andrew, A.S., Karagas, M.R., Hamilton, J.W., 2003. Decreased DNA repair geneexpression among individuals exposed to arsenic in united statesdrinking water. International Journal of Cancer 104 (3), 263–268.

Appelo, C.J., Van Der Weiden, M.J., Tournassat, C., Charlet, L., 2002. Surfacecomplexation of ferrous iron and carbonate on ferrihydrite and themobilization of arsenic. Environmental Science&Technology 36, 3096–3103.

ATSDR, 2003. Public Health evaluation of arsenic-contaminated privatedrinking water wells in Northeastern Connecticut Woodstock, EastHampton and Colchester Counties, Connecticut. US Dept. Health andHuman Services, Agency for Toxic Substances and Disease Registry.

Aurilio, A.C., Durant, J.L., Hemond, H.F., Knox, M.L., 1995. Sources anddistribution of arsenic in the Aberjona Watershed, Eastern Massachu-setts. Water Air and Soil Pollution 81 (3–4), 265–282.

Ayotte, J.D., et al., 2006. Bladder cancer mortality and private well use in NewEngland: an ecological study. Journal of Epidemiology and CommunityHealth 60 (2), 168–172.

Ayotte, J.D., Neilsen, M.G., Robinson, G.R., Moore, R.B., 1999. Relation ofarsenic, iron, and manganese in ground water to aquifer type, bedrocklithogeochemistry, and land use in the New England Coastal Basins.Water Resources Investigations Report #99-4162. United States Geolo-gical Survey, Pembroke, NH.

Ayotte, J.D., Montgomery, D.L., Flanagan, S.M., Robinson, K.W., 2003. Arsenicin Groundwater in Eastern New England: occurrence, controls, andhuman health implications. Environmental Science & Technology 37 (10),2075–2083.

Barnard, N.Q., 2006. A Geochemical, Petrologic, and Hydrologic Investigationof Two Bedrock Arsenic Sources, South-Central Maine. B.A. Thesis,Middlebury College, Middlebury, VT, 79 pp.

Boudette, E.L., et al., 1985. High Levels of Arsenic in the Groundwaters ofSoutheastern New Hampshire: A Geochemical Reconnaissance. UnitedStates Geological Survey, Concord, NH. Open File Report #85-202.

Burkert, L., 2005. Geochemical Controls on Arsenic Distribution andMobilization in Groundwater of Southeastern Pennsylvania. M.S. Thesis,Lehigh University, Bethlehem, PA, 58 pp.

Chang, T.K., Shyu, G.S., Lin, Y.P., Chang, N.C., 1999. Geostatistical, analysis ofsoil arsenic content in Taiwan. Journal of Environmental Science andHealth Part a-Toxic/Hazardous Substances & Environmental Engineering34 (7), 1485–1501.

Colt, J.S., et al., 2002. Sampling privatewells at past homes to estimate arsenicexposure: a methodologic study in New England. Journal of ExposureAnalysis and Environmental Epidemiology 12 (5), 329–334.

Davis, A., Kempton, J.H., Nicholson, A., Yare, B., 1994. Groundwater transportof arsenic and chromium at a Historical Tannery, Woburn, Massachu-setts, USA. Applied Geochemistry 9 (5), 569–582.

deLemos, J.L., Bostick, B.C., Renshaw, C.E., Sturup, S., Feng, X.H., 2006. Landfill-stimulated iron reduction and arsenic release at the Coakley SuperfundSite (Nh). Environmental Science & Technology 40 (1), 67–73.

Faill, R.T., 1973. Tectonic development of the Triassic Newark–GettysburgBasin in Pennsylvania. Geological Society of America Bulletin 84,725–740.

Faill, R.T., 1997a. A geologic history of the North-Central Appalachians.1.Orogenesis from the Mesoproterozoic through the Taconic Orogeny.American Journal of Science 297 (6), 551–619.

Faill, R.T., 1997b. A geologic history of the North-Central Appalachians.2. TheAppalachian Basin from the Silurian through the Carboniferous. Amer-ican Journal of Science 297 (7), 729–761.

Guidotti, C.V., Teichmann, F., Henry, D.J., 1991. Chlorite-bearing polymeta-morphic metapelites in the Rangeley Area, Maine — evidence forequilibrium assemblages. American Mineralogist 76 (5–6), 867–879.

Hatch, N.L., et al., 1984. Lithotectonic assemblages as portrayed on theNew Bedrock geologic map of Massachusetts. American Journal ofScience 284 (9), 1026–1034.

Hemond, H.F., 1995. Movement and distribution of arsenic in the AberjonaWatershed. Environmental Health Perspectives 103, 35–40.

Hitchcock, C., 1878. The Geology of New Hampshire: Part V. EconomicGeology. State Printer, Concord, New Hampshire. 98 pp.

Hohn, R., et al., 2006. Tracer test with as(V) under variable redox conditionscontrolling arsenic transport in the presence of elevated ferrous ironconcentrations. Journal of Contaminant Hydrology 88 (1–2), 36–54.

Hussey, A.M.I. and Berry, H.N.I., 2002. Bedrock Geology of the Bath 1:100,00Map Sheet, Coastal Maine, Maine Geological Survey, Bulletin #42.

Karagas, M.R., et al., 1996. Toenail samples as an indicator of drinking waterarsenic exposure. Cancer Epidemiology Biomarkers & Prevention 5 (10),849–852.

Karagas, M.R., et al., 1998. Design of an epidemiologic study of drinking waterarsenic exposure and skin and bladder cancer risk in a US population.Environmental Health Perspectives 106, 1047–1050.

Karagas, M.R., et al., 2001. Skin cancer risk in relation to toenail arsenicconcentrations in a US population-based case-control study. AmericanJournal of Epidemiology 153 (6), 559–565.

Karagas, M.R., et al., 2004. Incidence of transitional cell carcinoma of thebladder and arsenic exposure in New Hampshire. Cancer Causes &Control 15 (5), 465–472.

Keimowitz, A.R., et al., 2005a. Oxidation of groundwater arsenic and iron.Advances in Arsenic Research. Acs Symposium Series, pp. 206–219.

Keimowitz, A.R., et al., 2005b. Naturally occurring arsenic: mobilization at alandfill in Maine and implications for remediation. Applied Geochem-istry 20 (11), 1985–2002.

Keimowitz, A.R., et al., 2005c. Arsenic redistribution between sediments andwater near a highly contaminated source. Environmental Science &Technology 39 (22), 8606–8613.

Keppie, J.D., 1993. Synthesis of paleozoic deformational events and terraneaccretion in the Canadian Appalachians. Geologische Rundschau 82 (3),381–431.

Kim, M.-J., Nriagu, J., Hack, S., 2000. Carbonate ions and arsenic dissolution bygroundwater. Environmental Science and Technology 34 (15), 3094–3100.

Klaue, B., Blum, J.D., 1999. Trace analysis of arsenic in drinking water byinductively coupled plasma mass spectrometry: high resolution versushydride generation. Analytical Chemistry 71 (7), 1408–1414.

Knauer, K., Nepf, H.M., Hemond, H.F., 2000. The production of chemicalheterogeneity in UpperMystic Lake. Limnology AndOceanography 45 (7),1647–1654.

Kodama, K.P., Cioppa, M.T., Sherwood, E., Warnock, A.C., 1994. Paleomagnet-ism of baked sedimentary-rocks in the Newark and Culpeper Basins —

evidence for the J1-Cusp and Significant Late Triassic apparent polarwander from the Mesozoic basins of North-America. Tectonics 13 (4),917–928.

Kondo, H., et al., 1999. Naturally Occurring arsenic in the groundwaters in thesouthern region of Fukuoka Prefecture, Japan. Water Research 33 (8),1967–1972.

Lackovic, J.A., Nikolaidis, N.P., Dobbs, G.M.,1999. redox sensitivemobility of arsenicin proximity to a municipal landfill, hazardous and industrial wastes.Proceedings of the Mid Atlantic Industrial Waste Conference, pp. 451–459.

Lipfert, G., Reeve, A.S., Sidle, W.C., Marvinney, R., 2006. Geochemical patternsof arsenic-enriched ground water in fractured, crystalline bedrock,Northport, Maine, USA. Applied Geochemistry 21 (3), 528–545.

MacRae, J.D., Lavine, I.N., McCaffery, K.A., Ricupero, K., 2007. Isolation andcharacterization of Np4, arsenate-reducing Sulfurospirillum, from Mainegroundwater. Journal of Environmental Engineering-Asce 133 (1), 81–88.

Maguire, T.J., Sheridan, R.E., Volkert, R.A., 2004. Geophysical modeling of theNorthern Appalachian Brompton-Cameron, Central Maine, and Avalonterranes under the New Jersey coastal plain. Journal of Geodynamics 37(3–5), 457–485.

Maguire, T.J., Sheridan, R.E., Volkert, R.A., Feigenson, M.D., Patino, L.C., 1999.Continuation of Appalachian Piedmont under New Jersey coastal plain.In: Valentino, D.W., Gates, A.E. (Eds.), The Mid-Atlantic Piedmont:Tectonic Missing Link of the Appalachians. Special Paper, vol. 330.Geological Society of America, Boulder, CO.

Marshak, S., 2001. Earth: Portrait of a Planet. W. W. Norton and Company,W. W. Norton and Company, New York.

Marvinney, R.G., Loiselle, M.C., Hopeck, J.T., Braley, D., Krueger, J.A., 1994.Arsenic. Maine Groundwater: An Example from Buxton, Maine, FocusConference on Eastern Regional Ground Water Issues. National GroundWater Association, Burlington, VT, pp. 701–715.

Mayo, M.J., 2006. Arsenic Pathways in Landfills: A Case Study of the ShepleysHill Landfill, Ayer, Massachusetts. M.S. Thesis, Boston College, Boston,MA, 188 pp.

Mayo, M.J., Hon, R., Brandon, W.C., Ford, R., 2003. Arsenic in groundwater atlandfill sites in Northern Central Massachusetts. Abstracts of Papers ofthe American Chemical Society 226, U594.

Myers, T., Stewart, G., 1956. The geology of New Hampshire: Part III, mineralsand mines. New Hampshire State Planning and Development Commis-sion, Concord, New Hampshire. 107 pp.

New York Times, 1981a. Tests Planned in New Hampshire on Infants forArsenic in Water, New York Times, February 20, 1981, New York.

New York Times,1981b. A Town Is Troubled by ToxicWells and Pits, New YorkTimes, August 23, 1981, New York.

Nickson, R., MacArthur, J., Ravenscroft, P., Burgess, W., Ahmed, K., 2000.Mechanism of arsenic release to groundwater, Bangladesh and WestBengal. Applied Geochemistry 15, 403–413.

Nikolaidis, N.P., Dobbs, G.M., Chen, J., Lackovic, J.A., 2004. Arsenic mobility incontaminated lake sediments. Environmental Pollution 129 (3), 479–487.

Oen, I.S., Burke, E.A.J., Kieft, C., Westerhof, A.B., 1972.Westerveldite, (Fe,Ni,Co)as, a new mineral from La Gallega, Spain. American Mineralogist 57,354–363.

Olsen, P.E., Kent, D.V., Cornet, B., Witte, W.K., Schlische, R.W., 1996. High-resolution stratigraphy of the Newark Rift Basin (Early Mesozoic, EasternNorth America). Geological Society of America Bulletin 108 (1), 40–77.

Peters, S.C., Blum, J.D., 2003. The source and transport of arsenic in a bedrockaquifer, New Hampshire, USA. Applied Geochemistry 18 (11), 1773–1787.

21S.C. Peters / Journal of Contaminant Hydrology 99 (2008) 8–21

Peters, S.C., Burkert, L., 2007. The occurrence and geochemistry of arsenic ingroundwaters of the Newark Basin of Pennsylvania. Applied Geochem-istry 23, 85–98.

Peters, S.C., Blum, J.D., Klaue, B., Karagas, M.R., 1999. Arsenic occurrence in NewHampshire drinking water. Environmental Science & Technology 33 (9),1328–1333.

Peters, S.C., Blum, J.D., Karagas, M.R., Chamberlain, C.P., Sjostrom, D.J., 2006.Sources and exposure of the New Hampshire population to arsenic inpublic and private drinking water supplies. Chemical Geology 228 (1–3),72–84.

Pichler, T., Veizer, J., 1999. Precipitation of Fe(III) oxyhydroxide deposits fromshallow-water hydrothermal fluids in Tutum Bay, Ambitle Island, PapuaNew Guinea. Chemical Geology 162 (1), 15–31.

Pichler, T., Veizer, J., Hall, G.E.M., 1999. Natural input of arsenic into a coralreef ecosystem by hydrothermal fluids and its removal by Fe(III)oxyhydroxides. Environmental Science & Technology 33 (9), 1373–1378.

Price, R.E., Pichler, T., 2005. Distribution, speciation and bioavailability ofarsenic in a shallow-water submarine hydrothermal system, Tutum Bay,Ambitle Island, PNG. Chemical Geology 224 (1–3), 122–135.

Robinson, G.R., Ayuso, R.A., 2004. Use of spatial statistics and isotopic tracersto measure the influence of arsenical pesticide use on stream sedimentchemistry in New England, USA. Applied Geochemistry 19 (7),1097–1110.

Ryan, C.C., 1979. Lagoon of Arsenic Discovered in North Woburn, The DailyTimes, Woburn, MA.

Schlische, R.W., 1992. Structural and stratigraphic development of theNewark extensional basin, Eastern North-America — evidence for thegrowth of the basin and its bounding structures. Geological Society ofAmerica Bulletin 104 (10), 1246–1263.

Schlische, R.W., 2003. Progress in understanding the structural geology, basinevolution, and tectonic history of the Eastern North American RiftSystem. In: Le Tourneau, P.M., Olsen, P.E. (Eds.), The Great Rift Valleys ofPangea. Columbia University Press, New York.

Senn, D.B., Hemond, H.F., 2002. Nitrate controls on iron and arsenic in anUrban Lake. Science 296 (5577), 2373–2376.

Serfes, M., 2005. Arsenic Occurrence, Sources, Mobilization, Transport andPrediction in the Major Bedrock Aquifers of the Newark Basin. Ph.D.Thesis, Rutgers University, 122 pp.

Serfes, M.E., Spayd, S.E., Herman, G.C., 2005. Arsenic occurrence, sources,mobilization, and transport in groundwater in the Newark Basin of NewJersey,. Advances in Arsenic Research. Acs Symposium Series, pp. 175–190.

Sidle, W.C., 2002. O-18(So4) and O-18(H2o) as prospective indicators ofelevated arsenic in the Goose River ground-watershed, Maine. Environ-mental Geology 42 (4), 350–359.

Sidle, W.C., 2003. Identifying discharge zones of arsenic in the Goose RiverBasin,Maine. Journal of the AmericanWater Resources Association 39 (5),1067–1077.

Sidle, W.C., Fischer, R.A., 2003. Detection of H-3 and Kr-85 in groundwaterfrom arsenic-bearing crystalline bedrock of the Goose River Basin, Maine.Environmental Geology 44 (7), 781–789.

Sidle, W.C., Wotten, B., Murphy, E., 2001. Provenance of geogenic arsenic inthe Goose River Basin, Maine, USA. Environmental Geology 41 (1–2),62–73.

Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behavior anddistribution of arsenic in natural waters. Applied Geochemistry 17 (5),517–568.

Spear, F.S., Kohn,M.J., Cheney, J.T., Florence, F., 2002.Metamorphic, thermal, andtectonic evolution of Central New England. Journal of Petrology 43 (11),2097–2120.

Stollenwerk, K.G., Colman, J.A., 2002. Natural remediation potential ofarsenic-contaminated ground water In: Welch, A.H., Stollenwerk, K.G.(Eds.), Arsenic in Ground Water: Geochemistry and Occurrence. Kluwer,pp. 351–381.

Stollenwerk, K.G., Colman, J.A., 2004. Natural Remediation of ArsenicContaminated Ground Water Associated with Landfill Leachate. UnitedStates Geological Survey, Denver, CO. Fact Sheet 2004–3057.

Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry: Chemical Equilibria andRates in Natural Waters. Environmental Science and Technology. Wiley-Interscience, New York. 1022 pp.

Tseng, C.H., Chong, C.K., Chen, C.J., Tai, T.Y., 1996. Dose–response relationshipbetween peripheral vascular disease and ingested inorganic arsenicamong residents in Blackfoot disease endemic villages in Taiwan.Atherosclerosis 120 (1–2), 125–133.

Tucker, R.D., Osberg, P.H., Berry, H.N.I., 2001. The geology of part of Acadia andthe Nature of the Acadian Orogeny across Central and Eastern Maine.American Journal of Science 301, 205–260.

US-EPA, 1981. Investigation of Arsenic Sources in Groundwater. Environ-mental Protection Agency, Region 1, Boston, MA.

Utsunomiya, S., Peters, S.C., Blum, J.D., Ewing, R.C., 2003. Nanoscalemineralogy of arsenic in a region of New Hampshire with elevated as-concentrations in the groundwater. American Mineralogist 88 (11–12),1844–1852.

Wang, S., Mulligan, C.N., 2006. Occurrence of arsenic contamination inCanada: sources, behavior and distribution. Science of the TotalEnvironment 366, 701–721.

Welch, A.H., Lico, M.S., 1998. Factors controlling arsenic and uranium inshallow ground water, Southern Carson Desert, Nevada. AppliedGeochemistry 13 (4), 529–539.

Welch, A.H., Westjohn, D.B., Helsel, D.R., Wanty, R.B., 2000. Arsenic in groundwater of the United States: occurrence and geochemistry. Ground Water38 (4), 589–604.

Weldon, J.M., MacRae, J.D., 2006. Correlations between arsenic in Mainegroundwater and microbial populations as determined by fluorescencein situ hybridization. Chemosphere 63 (3), 440–448.

West, D.P., Yates, M.G., Gerbi, C., Barnard, N.Q., 2008. MetamorphosedOrdovician iron- and manganese-rich rocks in South-Central Maine:from Peri-Gondwanan deposition through Acadian metamorphism.American Mineralogist 93, 270–273.

Witte,W.K., Kent, D.V., 1991. Tectonic implications of a remagnetization eventin the Newark Basin. Journal of Geophysical Research-Solid Earth 96(B12), 19569–19582.

Zuena, A.J., Keane, N.W.,1985. Arsenic contamination of private potablewells.US EPA National Conference on Environmental Engineering, North-eastern University, Boston, MA, pp. 717–725.