, 1123 (2010);
et al.Scott Fendorf,
Arsenic in South and Southeast Asia
Spatial and Temporal Variations of Groundwater
This copy is for your personal, non-commercial use only.
. clicking herecolleagues, clients, or customers by
, you can order high-quality copies for your
If you wish to distribute this article to others
. herefollowing the guidelines
can be obtained by
Permission to republish or repurpose articles or portions of articles
(this information is current as of June 4, 2010 ):
The following resources related to this article are available online at www.sciencemag.org
version of this article at:
including high-resolution figures, can be found in the online
Updated information and services,
can be found at:
Supporting Online Material
, 7 of which can be accessed for free:
cites 50 articles
This article appears in the following
registered trademark of AAAS.
is aScience2010 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on June 4, 2010
Spatial and Temporal Variations of
Groundwater Arsenic in South and
Scott Fendorf,1* Holly A. Michael,2* Alexander van Geen3*
Over the past few decades, groundwater wells installed in rural areas throughout the major river basins
draining the Himalayas have become the main source of drinking water for tens of millions of people.
Groundwater in this region is much less likely to contain microbial pathogens than surface water but
from rocks and sediment by coupled biogeochemical and hydrologic processes, some of which are
presently affected by human activity. Mitigation of the resulting health crisis in South and Southeast
Asia requires an understanding of the transport of arsenic and key reactants such as organic carbon that
could trigger release in zones with presently low groundwater arsenic levels.
evated groundwater arsenic (As) concentrations
in many parts of Bangladesh. Estimates of the
rural population exposed to unsafe As levels
by drinking untreated groundwater in India,
China, Myanmar, Pakistan, Vietnam, Nepal, and
Cambodia have grown to over 100 million (2).
and epidemiological studies conducted elsewhere
lead to predictions of a doubling of the lifetime
der, and lung (3, 4). Groundwater containing As
mental development of children (5, 6).
The affected areas of South and Southeast
of rivers that drain the Himalayas (Fig. 1A) (7).
host increasing numbers of inexpensive wells
made of polyvinyl chloride pipe with a cast-iron
handpump mounted on top (tubewells) that are
installed to avoid drinking surface water con-
though by no means sufficient, testing of tubewell
water for As has been carried out in most of the
Southeast Asia there is extensive variation in the
depth distribution of wells (Fig. 1B). In Bangla-
desh and the bordering state of West Bengal,
he largest poisoning of a population
in history” is how Smith et al. (1)
described the health impact of el-
India, tubewells extend to depths of ~350 m com-
pared to a maximum of ~100 m in Nepal, Cam-
bodia, and Vietnam, owing to difference in the
thickness of unconsolidated sand deposits (8).
More than half the wells in at least one depth
interval in each of the five affected countries do
not meet the World Health Organization (WHO)
guideline of 10 mg/liter As in drinking water
(Fig. 1B). There are also numerous wells con-
taining <10 mg/liter As at all depths. The ex-
tensive spatial variability of As concentrations at
hinders comparisons among field sites and the rec-
ognition of presumably common biogeochemical-
hydrological processes that regulate As levels in
groundwater. The source of As is not a mystery,
however; what is less clear is how the current dis-
tribution of dissolved As in the subsurface reached
its current state. This review focuses on what has
been learned from a decade of field research
conducted in South and Southeast Asia about
the processes that resulted in the current dis-
tribution of As in groundwater and the key
factors that will control changes in the distri-
bution of As over time.
What Drives the Release of
Arsenic to Groundwater?
Weathering of Himalayan-derived sediment dur-
ing erosion and transport leads to downstream dep-
osition of As. The primary sources of As within
the Himalayas are thought to be eroding coal
Exposed to the atmosphere, the minerals con-
tained within these deposits are oxidized, and
much of their As content is transferred to sec-
ondary phases including iron (Fe) hydroxides,
oxyhydroxides, and oxides, collectively referred
a positive relation between As and Fe extracted
from hundreds of sediment samples from the
Ganges-Brahmaputra-Meghna, Mekong, and Red
River basins (Fig. 2) (15–18). Grain-size separa-
tion of river-borne and aquifer sedi-
high–surface area fraction (<10 mm)
now recognized as a key step in the
water, with other phases possibly
into groundwater as a result of two
der the anoxic conditions that pre-
laboratory evidence suggest that mi-
crobial reduction of Fe(III) oxides
liberates As into the dissolved phase
labile As(III) probably contributes to
this release but is hard to distinguish
from the reduction of Fe oxides under natural
conditions given the rates of groundwater flow.
Second, dissolution of Fe oxides is accompanied
by the release of other ligands such as phosphate
that compete with As for adsorption on the
remaining Fe oxide surface sites (9).
The restriction of high dissolved As concen-
indicative of coatings of reduced or mixed-valence
Fe(II+III) oxides, and the absence of elevated con-
centrations from aquifers containing orange sands
coated with Fe(III) oxides (Box 1) suggest that Fe
(III) reduction is a primary factor contributing to
high As concentrationsin groundwater (9, 24–28).
A systematic analysis of the composition of hun-
dreds of groundwater samples from the Bengal,
Mekong, and Red River basins has shown that
high concentrations of As in groundwater prevail
under advanced stages of reduction rather than
the onset of Fe oxide reduction (29).
Microbial Fe(III) and As(V) reduction both
require a supply of labile organic carbon. When
the biological oxygen demand from the decom-
1Department of Earth System Sciences, Stanford University,
*To whom correspondence should be addressed. E-mail:
email@example.com (S.F); firstname.lastname@example.org (H.A.M.);
CREDIT: SCOTT FENDORF
VOL 328 28 MAY 2010
on June 4, 2010
position of organic carbon exceeds the rate of ox-
as well as As(V) to As(III). Elevated groundwater
As concentrations that broadly correspond with
increased levels of metabolic by-products in
groundwater including inorganic carbon, ammo-
are consistent with the central role of organic-
matter metabolism (18, 19, 28–30).
Where Does Arsenic Release to
There are three environmental requirements for
saturation (which limits diffusion of atmospheric
oxygen), a limited supply of sulfur, and a source
of organic carbon to drive microbial dissolution
of Fe oxides. The height of the water table, typ-
ically within 5 m of the surface, indicates where
oxygen supply is limited and reductive dissolu-
tion can potentially be initiated (Fig. 3B). The
domain within which As can be released to
aquifers where sulfate supplied by recharge has
not been depleted. This is because sulfate re-
duction promoted by organic carbon produces
sulfide that can bind As, forming sparingly sol-
uble sulfides mineral that effectively remove As
from groundwater (13, 29). Marine-influenced
areas also show inhibition of As release by sul-
fate reduction along the coasts of Bangladesh (9)
and Vietnam (31).
The availability of labile organic carbon as a
driver of microbial reduction is possibly the most
prominent outstanding issue limiting our ability
to predict the distribution of As in groundwater.
Organic carbon necessary to drive reduction of
pathways. One is co-deposition of plant material
as an autochthonous source of carbon (9). Dis-
solved organic carbon (DOC), produced by re-
cent degradation of plants in modern soils or in
buried peat layers and transported to a different
location by groundwater flow, could be an alter-
native allochtonous supply (26, 28). The reac-
tivity of organic matter needs to be considered as
well (32, 33), as indicated by dissolved inorganic
and by assays of microbial decomposition (34).
The relative importance of different sources of
organic carbon remains undetermined and even
sediment and in what quantity will depend on the
of As in the sediment. Sediment with recalcitrant
organic carbon and/or As-bearing Fe oxides is
expected to release As slowly. In contrast, highly
reactive forms of both organic carbon and labile
sediment-bound As should result in the strongest
release. Field evidence from Nepal, West Bengal,
Bangladesh, Cambodia, and Vietnam suggests
both rapid, shallow release of As as well as more
The available data show that the geological
setting likely plays an important role, but there
Nepal India (West Bengal)BangladeshCambodia Vietnam
Fraction of wells with >10 g/liter per depth quartile
Well water As ( g/liter)
0 400800 km
0.90.00.30.6 0.90.00.3 0.60.9 0.00.30.6 0.90.00.30.60.9
1 10 100 1000 110 100100 1000 1101001000 110 100 1000
Fig. 1. Distribution of arsenic in groundwater of South and Southeast Asia. (A)
Map of four major river basins draining the Himalayas. (B) Depth distribution of
As in groundwater determined for five affected countries. Concentrations of As
are shown on a logarithmic scale. Symbols are color-coded according to the major
river basins shown in (A). The pink line depicts the fraction of wells that exceed
28 MAY 2010VOL 328
on June 4, 2010
remain notable uncertainties regard-
ing rates of carbon metabolism cou-
pled to As release.
The pool of labile As within an
interval of an aquifer sediment is fi-
continued reduction of Fe oxides.
Such a situation has been docu-
mented for deeper aquifers of Bang-
ladesh where dissolved As levels are
low despite elevated Fe(II) concen-
trations in groundwater (16). In other
situations, the available pool of labile
organic carbon has been depleted al-
though some labile As is still bound
to sediment particles. Sediments de-
posited prior to about 20,000 years
ago and that were well drained be-
sea-level low stand, for instance, con-
tain limited reactive organic matter.
The orange color of these oxidized
deposits indicates that they were de-
posited with a low concentration of
organic carbon or that their initial
organic carbon was oxidized during the low stand
(9, 15, 16, 25, 26).
After the initial biogeochemical transforma-
tions that result in As release from the sediment,
adsorption on residual or newly formed aquifer
solids will control dissolved As concentrations.
Weaker surface complexes of As(III) and the
degradation of Fe oxides (9) mean that adsorption
is less pronounced than for As(V) in oxidized
surface environments (37). Nevertheless, adsorp-
tion of As(III) does occur within reduced aquifers,
conditions in Bangladesh (38). This implies that
As transport is substantially retarded relative to
groundwater flow, even if adsorption sites may be
saturated in aquifer sands under certain conditions
How Does Groundwater Flow Affect the
Distribution of Arsenic?
Groundwater flow transports dissolved As as well
as DOC, oxygen, sulfate, and competing adsorb-
ates, all of which influence As concentrations.
When the system is not in a steady state, either
hydrologically or biogeochemically, As concen-
trations can be expected to change over time.
current distribution of groundwater As and its
The main river basins affected
by As (Fig. 1A) share similar hy-
drogeologic features, most notably
a monsoonal climate and rapid
sediment accumulation. Ground-
water flow systems range in scale
from the local (tens of meters) to
the regional (hundreds of kilo-
systems (39–42), which are most
relevant to the distribution of As in
plex, site-specific, and transient na-
ture of natural patterns of recharge
and discharge (Fig. 3B). Further,
abundant surface water bodies such
as rivers, ponds, and wetlands inter-
act with the groundwater systems.
Monsoonal rains and dry-season
irrigation pumping cause reversals
in hydraulic gradients that can trans-
a sink of groundwater and back
over a year (39–42). Constructed
in the Bengal Basin and vary in their contribu-
tion to aquifer recharge (34, 39, 43), depending
on the accumulation of fine-grained bottom
sediment. Such seasonally and spatially varia-
ble forcing can result in highly complex ground-
High groundwater pumping can substantially
of groundwater pumping for irrigation is at least
an order of magnitude higher than integrated
flow from hand pumps (16, 39, 44). Irrigation
recharge and discharge areas, increase recharge
greatest in the Bengal Basin and the Terai Basin
along the southern border of Nepal, less in the
Red River Basin (17), and least in the Mekong
River Basin (41). Because elevated As concen-
trations are observed in all these areas (Fig. 1B),
processes associated with irrigation pumping,
though potentially important, cannot be the only
trigger of As release to groundwater.
The time since recharge, or groundwater age,
is also an important factor that influences ground-
water As concentrations. Groundwater age, mea-
sured by two different radioactive clocks, ranges
from less than 1 year to a few decades in shallow
The vertical gradient in groundwater ages reflects
regional flow systems and flowpaths that link
distant recharge and discharge areas beneath more
vigorous shallow and local groundwater circula-
tion (Fig. 3B). Irrigation water is typically drawn
from shallow (<100 m) depths and may be partly
responsible for the pronounced difference in age
between shallow and deeper aquifers (39, 45).
Extractable As (mg/kg)
Extractable Fe (g/kg)
Fig. 2. RelationbetweenAsandFeconcentrationsforasuiteofsedimentsamples
from three countries based on different extraction methods (9, 14, 18, 50).
Box 1: The color of aquifer
sands is a useful visual indicator
of the redox state of an aquifer.
(Bottom) Orange sands from
Vietnam indicate the presence
of Fe(III) oxides that are
consistently associated with
low-As water, whereas gray
sands with reduced or mixed-
valence Fe(II+III) oxides (top)
are often, though not always,
associated with higher dissolved
As. Sand color has been used by
drillers to target low-As
groundwater in spatially
heterogeneous aquifers. [Photo
courtesy of Benjamin Bostick]
VOL 328 28 MAY 2010
on June 4, 2010
Within the river basins considered here, the
water in shallow strata may be due to differences
in topography on multiple scales. Slightly ele-
vated, often coarse, sandy deposits appear to be
associated with lower As concentrations in
Bangladesh and Cambodia (47–49). Such obser-
vations suggest that rapid recharge through these
deposits locally inhibits the release of As, pos-
sibly by supplying oxygen, nitrate, or sulfate as
alternatives to Fe oxides for oxidizing organic
carbon (34,48).Similar processes prevent release
of arsenic in water recharged through rice field
bunds (34). In contrast, low-lying areas in the
sediment, frequently flooded, and associated with
Rapid release of As under these conditions is
attributed to co-deposition of labile carbon and
As-bearing Fe oxides in the seasonally saturated
surface sediments (18, 30), infiltration of recharge
with abundant DOC (17, 39), or simply slow flow
of water through As-releasing sediment (48).
Along the pathway of groundwater flow,
changes in As concentrationwill depend on local
partitioning (adsorption/desorption) with the sed-
releasedfrom the sedimentandeventuallyflushed
from the aquifer in areas where the concentration
of As in inflowing water is below that dictated by
partitioning, even within reduced gray sands de-
pleted in Fe(III) (38). Along anaerobic, shallow
flowpaths containing organic carbon, As concen-
correlation between As and groundwater age or
flow rate within shallow aquifers (21, 40), and
with As plumes that originate from natural wet-
lands high in organic carbon (18, 30) or con-
structed ponds (34). The subsurface maximum in
groundwater As frequently observed within shal-
low gray reduced aquifers is likely the result of
layering of groundwater flow having different
evolutionary histories. High-As groundwater indi-
dissolution/desorption; low-As water can reflect
flowpaths that lack Fe/As reduction, secondary
As-sulfide precipitation, various extents of sedi-
(16, 28, 34, 38, 45).
How Vulnerable Are Low-Arsenic Zones?
Low-As zones within the aquifer systems of the
be associated with reduced gray [Fe(II) domi-
nated] or oxidized orange [Fe(III) dominated]
sands (16). Zones of low dissolved As occur in
gray sands where As is removed by sulfide (13)
flushed by sustained recharge or has never been
released (16, 38). Low-As zones associated with
oxidized orange sands are often deeper (>100 m)
but, depending on the local geology, are occa-
Groundwater is typically anoxic throughout the
affected region, even within orange sand de-
posits. The vulnerability of shallower and deeper
low-As zones to human perturbation must be un-
derstood because millions of households in
Bangladesh have switched their consumption to
a nearby low-As well identified by testing in the
Low-As zones can be protected against in-
trusion of high-As groundwater by favorable
hydraulics or geochemical processes. Hydraulic
protection occurs where the source area that con-
tributes water to a particular zone is not high in
dissolved As or solutes that can mobilize As.
Geochemical protection occurs because of As ad-
Experiments and modeling indicate that break-
through of As through 10 m of orange sands may
lag groundwater flow by hundreds of years
because of adsorption (37).
Shallow low-As zones are particularly vul-
flow combined with the patchy distribution of
dissolved and solid-phase As (Fig. 3B). The
adsorption capacity of gray sands that prevail at
shallow depths is lower than that of orange sands
and further contributes to the vulnerability of
shallow low-As zones. A primary threat is ad-
vective transport from adjacent high-As zones
because groundwater flows much more easily
Deeper groundwater (>100m deep) is more
uniformly low in As (Fig. 1B) and already a
Transfer of As from
sulfide to Fe oxide
Release of As
from Fe oxides
Trapping of As
at oxic interface
Fe(III) + As(V)
Fe(II) + As(III)
Fe(II) + As(III)
As trapping by
Regional groundwater flow
As sorbs more strongly
to orange Fe(III) oxides
than to gray sediments
Fig. 3. (A) Depth distribution of groundwater ages in Bangladesh de-
termined by either the3H-3He method (red symbols) or radiocarbon in those
Concentrations of As were not reported for three deep samples shown as
3H was measured and not detected (blue symbols) (54).
light blue circles but are likely to be low; the age of samples shown as gray
circles is uncertain owing to their low3He content. (B) Conceptual diagram
28 MAY 2010VOL 328
on June 4, 2010
widely used and potentially sustainable source of
is lowin As.Deep,regionalsystems likely occurin
much of the Bengal Basin despite low regional
hydraulic gradients because of its large extent,
depth, and extreme vertical heterogeneity (44, 52).
Hydraulic gradients in the Mekong (41), Red River
(42), and Terai basins are similarly low, but the
more limited regional flow. Numerical modeling of
groundwater flow in the Bengal Basin (44) has
shownthathydraulicprotection may lastfor atleast
1000 years in much of the As-affected area if wells
aquifers are limited to domestic supply. In contrast,
deep pumping for irrigation occurs at order-of-
magnitude higher rates compared to hand pumps
and could induce much earlier and larger-scale
downward migration of As (44).
Human-induced changes have and will con-
tinue to threaten low-As zones. Whereas hydrau-
lic heads and flow velocities respond quickly to
changes in physical forcing, solute concentra-
tions require a period at least equivalent to the
groundwater residence time to reach a new equi-
such as irrigation pumping or the digging of
ponds, has been approximately equivalent to the
residence time of groundwater (39). Raising vil-
lages above flood level using low-permeability
clay also creates a cap that inhibits recharge and
could lead to a buildup of As in shallow aquifers
previously suitable for drinking (49). Transport
of reactive DOC by irrigation pumping into a
zone of either gray or orange sands that currently
the onset of reductive dissolution and As release
into groundwater (28, 34).
Where hydraulic protection does not exist,
in shallower strata or where pumping rates are
high, a zone may remain low in As for extended
periods because of retardation by adsorption (37).
The delay in the appearance of elevated levels of
As should decrease with increasing velocity and
increase with flow distance, particularly through
orange sands. The lower rate at which water is
drawn from a community hand pump is therefore
preferable to a mechanized pump connected to a
should extend as deep as possible into deep
orange sands. These recommendations are con-
sistent with the outcome of monitoring a set of
hand-pumped community wells in Bangladesh
were recorded during the initial years (mostly in
wells <60 m deep) and none since (53).
Priorities for the Future
The laterally and vertically heterogeneous dis-
tribution of As has one advantage—many vil-
lagers in the affected regions live within walking
distance of a well that is low in As or within
drilling distance of such a zone. Governments
and international organizations should therefore
kits. Better use should also be made of existing
geological data and compiling test results to target
those zones that are low in As for the installation
of community wells. Even if wells tapping deeper
strata are more likely to be hydraulically and
chemically protected, tens of thousands of deep
be retested, which currently happens rarely.
As outlined in this review, the downside of
patchiness is poor predictability and potential
sensitivity of those aquifers that are low in As to
changes in flow and/or related biogeochemical
should therefore be coupled with more detailed
evaluation of the local hydrology. Another topic
that deserves closer study is the viability of rural
some governments and international organiza-
anized pumps concentrate deep pumping and are
Every effort should be made to prevent irri-
gation by pumping from deeper aquifers that are
low in As. The accumulation of As in paddy soil
and rice grains is a source of concern, but deep
aquifers should not be compromised by abstrac-
route of exposure to As.
References and Notes
1. A. H. Smith, E. O. Lingas, Bull. World Health Organ. 78,
2. P. Ravenscroft, H. Brammer, K. Richards, Arsenic
Pollution: A Global Synthesis (RGS-IBG Book Series,
Wiley-Blackwell, Chichester, UK, 2009).
3. U. K. Chowdhury et al., Environ. Health Perspect. 108,
4. Y. Chen, H. Ahsan, Am. J. Public Health 94, 741 (2004).
5. C. J. Chen, H. Y. Chiou, M. H. Chiang, L. J. Lin, T. Y. Tai,
Arterioscler. Thromb. Vasc. Biol. 16, 504 (1996).
6. G. A. Wasserman et al., Environ. Health Perspect. 112,
7. L. Winkel, M. Berg, M. Amini, S. J. Hug, C. Annette
Johnson, Nat. Geosci. 1, 536 (2008).
8. S. L. Goodbred Jr., S. A. Kuehl, Sediment. Geol. 133, 227
9. D. G. Kinniburgh, P. L. Smedley, Eds., Arsenic
Contamination of Ground Water in Bangladesh, Final
Report (BGS Technical Report WC/00/19, British
Geological Survey, Keyworth, UK, 2001), vol. 2.
10. A. van Geen et al., Water Resour. Res. 39, 1140 (2003).
11. R. Nickson et al., J. Environ. Sci. Health Part A Tox.
Hazard. Subst. Environ. Eng. 42, 1707 (2007).
12. S. K. Acharyya et al., Nature 401, 545, discussion 546
13. H. A. Lowers et al., Geochim. Cosmochim. Acta 71, 2699
14. B. D. Kocar et al., Appl. Geochem. 23, 3059 (2008).
15. C. H. Swartz et al., Geochim. Cosmochim. Acta 68, 4539
16. Y. Zheng et al., Geochim. Cosmochim. Acta 69, 5203 (2005).
17. M. Berg et al., Chem. Geol. 249, 91 (2008).
18. D. Postma et al., Geochim. Cosmochim. Acta 71, 5054
19. C. B. Dowling et al., Water Resour. Res. 38, 1173 (2002).
20. A. A. Seddique et al., Appl. Geochem. 23, 2236 (2008).
21. B. Nath et al., J. Hydrol. (Amst.) 364, 236 (2009).
22. B. J. Mailloux et al., Appl. Environ. Microbiol. 75, 2558
23. F. S. Islam et al., Nature 430, 68 (2004).
24. K. J. Tufano, C. Reyes, C. W. Saltikov, S. Fendorf, Environ.
Sci. Technol. 42, 8283 (2008).
25. A. Horneman et al., Geochim. Cosmochim. Acta 68, 3459
26. J. M. McArthur et al., Water Resour. Res. 44, W11411
27. M. von Brömssen et al., Sci. Total Environ. 379, 121
28. C. F. Harvey et al., Science 298, 1602 (2002).
29. J. Buschmann, M. Berg, Appl. Geochem. 24, 1278 (2009).
30. M. L. Polizzotto, B. D. Kocar, S. G. Benner, M. Sampson,
S. Fendorf, Nature 454, 505 (2008).
31. S. Jessen et al., Appl. Geochem. 23, 3116 (2008).
32. H. A. L. Rowland et al., Geobiology 5, 281 (2007).
33. N. Mladenov et al., Environ. Sci. Technol. 44, 123
34. R. B. Neumann et al., Nat. Geosci. 3, 46 (2010).
35. J. Gurung, H. Ishiga, M. S. Khadka, Environ. Geol. 49, 98
36. N. C. Papacostas, B. C. Bostick, A. N. Quicksall,
J. D. Landis, M. Sampson, Geology 36, 891 (2008).
37. K. G. Stollenwerk et al., Sci. Total Environ. 379, 133
38. A. van Geen et al., Environ. Sci. Technol. 42, 2283 (2008).
39. C. F. Harvey et al., Chem. Geol. 228, 112 (2006).
40. M. Stute et al., Water Resour. Res. 43, W09417 (2007).
41. S. G. Benner et al., Appl. Geochem. 23, 3072 (2008).
42. F. Larsen et al., Appl. Geochem. 23, 3099 (2008).
43. S. Sengupta et al., Environ. Sci. Technol. 42, 5156
44. H. A. Michael, C. I. Voss, Proc. Natl. Acad. Sci. U.S.A.
105, 8531 (2008).
45. S. Klump et al., Environ. Sci. Technol. 40, 243 (2006).
46. A. Mukherjee, A. E. Fryar, P. D. Howell, Hydrogeol. J. 15,
47. J. Buschmann, M. Berg, C. Stengel, M. L. Sampson,
Environ. Sci. Technol. 41, 2146 (2007).
48. Z. Aziz et al., Water Resour. Res. 44, W07416 (2008).
49. B. Weinman et al., Geol. Soc. Am. Bull. 120, 1567 (2008).
50. E. Eiche et al., Appl. Geochem. 23, 3143 (2008).
51. M. F. Ahmed et al., Science 314, 1687 (2006).
52. P. Ravenscroft, W. G. Burgess, K. M. Ahmed, M. Burren,
J. Perrin, Hydrogeology J. 13, 727 (2005).
53. A. van Geen et al., J. Environ. Sci. Health Part A Tox.
Hazard. Subst. Environ. Eng. 42, 1729 (2007).
54. Data sources and methods are available on Science
55. We thank J. W. Rosenboom of Water Supply Program/
World Bank and M. A. Sampson, founder and director
of RDI International, Cambodia (deceased 19 March
2009), for help in organizing the AGU Chapman
conference held in Siem Reap, Cambodia, in March 2009
that resulted in this review. Travel support for the
conference was contributed by the Woods Institute for the
Environment at Stanford University and the European
Union Asia-Link CALIBRE Project. We also acknowledge
research funding by the Environmental Venture Projects
program of Stanford’s Woods Institute for the
Environment and the Stanford NSF Environmental
Molecular Science Institute (NSF-CHE-0431425) (S.E.F.),
the U.S. Geological Survey, the U.S. Agency for
International Development, the British Department for
International Development, and UNICEF (H.A.M.),
National Institute of Environmental Health Sciences SRP
grant 1 P42 ES10349, NIH FIC grant 5 D43 TW05724,
and NSF grant EAR 0345688 (AvG). We thank R. Beckie,
G. Breit, C. Harvey, J. Lloyd, and an anonymous reviewer
for helpful comments. This is Lamont-Doherty Earth
Observatory contribution no. 7355.
VOL 32828 MAY 2010
on June 4, 2010