Behaviour of estrogenic endocrine-disrupting chemicals in permeable carbonate

RESEARCH ARTICLE
Behaviour of estrogenic endocrine-disrupting chemicals
in permeable carbonate sands
Benjamin O. Shepherd &Dirk V. Erler &Douglas R. Tait &
Lukas van Zwieten &Stephen Kimber &Bradley D. Eyre
Received: 16 September 2014 /Accepted: 13 February 2015
#Springer-Verlag Berlin Heidelberg 2015
Abstract The remediation of four estrogenic endocrine-
disrupting compounds (EDCs), estrone (E1), estradiol
(E2), ethinylestradiol (EE2) and estriol (E3), was mea-
sured in saturated and unsaturated carbonate sand-filled
columns dosed with wastewater from a sewage treatment
plant. The estrogen equivalency (EEQ) of inlet wastewa-
terwas1.2ngL
−1
and was remediated to an EEQ of
0.5ngL
−1
through the unsaturated carbonate sand-filled
columns. The high surface area of carbonate sand and
associated high microbial activity may have assisted the
degradation of these estrogens. The fully saturated sand
columns showed an increase in total estrogenic potency
withanEEQof2.4ngL
−1
, which was double that of the
inlet wastewater. There was a significant difference
(P<0.05) in total estrogenic potency between aerobic
and anaerobic columns. The breakdown of conjugated es-
trogens to estrogenic EDCs formed under long residence
timeandreducingconditionsmayhavebeenresponsible
for the increase in the fully saturated columns. This may
also be explained by the desorption of previously sorbed
estrogenic EDCs. The effect of additional filter materials,
such as basalt sediment and coconut fibre, on estrogenic
EDC reduction was also tested. None of these amend-
ments provided improvements in estrogen remediation
relative to the unamended unsaturated carbonate sand col-
umns. Aerobic carbonate sand filters have good potential
to be used as on-site wastewater treatment systems for the
reduction of estrogenic EDCs. However, the use of fully
saturated sand filters, which are used to promote denitri-
fication, and the loss of nitrogen as N
2
were shown to
cause an increase in EEQ. The potential for the accumu-
lation of estrogenic EDCs under anaerobic conditions
needs to be considered when designing on-site sand fil-
tration systems required to reduce nitrogen. Furthermore,
the accumulation of estrogens under anaerobic conditions
such as under soil absorption systems or leachate fields
has the potential to contaminate groundwater especially
when the water table levels fluctuate.
Keywords Estrogenic EDCs .Saturation state .Carbonate
sand .Remediation .Wastewater treatment systems .
Groundwater
Introduction
A growing area of environmental and human health concern is
contamination of groundwater with endocrine-disrupting
chemicals (EDCs) (Jenssen 2006; Kookana et al. 2007). The
United States Environmental Protection Agency (USEPA) de-
fines an EDC as Ban exogenous agent that interferes with the
synthesis, secretion, transport, binding, action or elimination
of natural hormones in the body that are responsible for the
maintenance of homeostasis, reproduction, development and/
Responsible editor: Hongwen Sun
B. O. Shepherd (*):D. V. Erler :D. R. Tait :B. D. Eyre
Centre for Coastal Biogeochemistry, School of Environment, Science
and Engineering, Southern Cross University, Lismore, NSW 2480,
Australia
e-mail: [e-mail]
L. van Zwieten :S. Kimber
Department of Primary Industries, Bruxner Highway,
Wollongbar, NSW 2477, Australia
L. van Zwieten
Southern Cross Plant Science, School of Environment, Science and
Engineering, Southern Cross University, Lismore, NSW 2480,
Australia
Environ Sci Pollut Res
DOI 10.1007/s11356-015-4238-8

or behaviour^. Chronic, low-level exposure to EDCs in both
humans and animals has been linked to a number of reproduc-
tive disorders (Kookana et al. 2007).
EDCs encompass a wide variety of chemical classes, in-
cluding natural and synthetic hormones, plant constituents,
pesticides, certain compounds in plastics, personal care prod-
ucts, pharmaceuticals, pollutants and by-products of industry
(Damstra et al. 2002). EDCs are widely dispersed in the envi-
ronment and have been found in rivers, estuaries, coastal sam-
ples and even in the arctic (Braga et al. 2005a;Jenssen2006).
Some of the most potent EDCs include steroid estrogens such
as estrone (E1), 17β-estradiol (E2), estriol (E3) and 17α-
ethinylestradiol (EE2). These estrogenic EDCs have been
shown to have adverse affects on the reproductive systems
of marine and freshwater animals at less than 5 ng L
−1
(Jobling
et al. 1998,2002a,b,2006; van Aerle et al. 2001; Pelley
2003). Estrone is a metabolite of estradiol and is an important
pregnancy estrogen. Estradiol is a reproductive hormone nat-
urally excreted by female animals including humans and in
much smaller quantities by males. Ethinylestradiol is the main
ingredient in the female contraceptive pill and hormone re-
placement medication (Trussell 2001;Songetal.2009).
A major source of estrogenic EDCs to natural waterways is
wastewater treatment plant effluent and leachate of sewage
sludge either buried in landfill or applied as fertilizer (Jobling
et al. 2006; Koyama et al. 2006; Thorpe et al. 2009;
Karnjanapiboonwong et al. 2011; Marti and Batista 2014).
Concentrations of estrogenic EDCs in wastewater can range
from nanograms per litre (ng L
−1
) to micrograms per litre (μg
L
−1
) (Ying and Kookana 2005). Other potential sources of
estrogenic EDCs in developed and agricultural catchments
include on-site wastewater management systems, effluent
run-off following manure application or run-off of livestock
wastes (Reneau and Degen 1989; Hanselman et al. 2003;
Zhang et al. 2011).
Water contamination by sewage is common in all island
countries of the Pacific region with many streams and ground-
water supplies currently considered unsafe for human con-
sumption without treatment (UNESCAP 2000). In the porous
carbonate sands of many Pacific islands, groundwater is par-
ticularly susceptible to contamination from poorly performing
sanitation systems. Indeed ground and surface water pollution
associated with septic tank and absorption trench disposal
systems is common in many Pacific countries (Dakers and
Evans 2007). Some soil absorption systems which are used
as on-site treatment systems in these countries can operate
successfully for years, whilst others can fail within weeks of
installation (Beal et al. 2005). Studies by
Karnjanapiboonwong (2011) showed that EDCs that were
present in the effluent from a wastewater treatment plant were
able to reach the groundwater after land application of the
effluent. Given the porous nature of the soil in most Pacific
islands, the steadily increasing human population and greater
use of synthetic compounds, such as the contraceptive pill, it
is quite likely that groundwater may contain elevated concen-
trations of estrogenic EDCs. However, it is also possible that
carbonate sands may act to physically, chemically and/or bio-
logically remove estrogenic EDCs from effluent water.
There is some evidence that EDCs can be effectively de-
graded during transport through porous media (Nakada et al.
2007). Biodegradation of EDCs has been shown to be the
most cost-effective way of removing EDCs relative to ad-
vanced removal mechanisms such as activated carbon, mem-
brane filtration, chemical advanced oxidation and ozonation
(Liu et al. 2009).Usingquartzsandfiltrationsystems,Nakada
et al. (2007) showed a 50 % removal rate of estrogenic EDCs.
Carbonate sands may be particularly effective for the treat-
ment of EDCs (Rasheed et al. 2003). They may be an ideal
candidate for incorporation into wastewater treatment systems
in developing Pacific nations due to their ready availability.
However, the behaviour of EDCs in carbonate sands has not
been studied previously. In particular, it is not understood how
the biodegradation of estrogens will be affected by the degree
of saturation within carbonate sands. Other studies have
shown that unsaturated aerobic soils are more likely to de-
grade EDCs relative to saturated anaerobic soils. Permeable
carbonate sands however are unique in that they have ex-
tremely high surface areas capable of sustaining a large mi-
crobial biomass and contain greater adsorption capabilities
and pH buffering relative to quartz sands (Rasheed et al.
2003; Schöttner et al. 2011). The behaviour and degradation
of EDCs in permeable carbonate sands therefore may be en-
hanced relative to other sands and soils. To date, there have
been no studies that have looked at the removal of EDCs in
carbonate sands under saturated or unsaturated conditions.
The aim of this study was to investigate the behaviour of
estrogenic compounds in permeable carbonate sands under
saturated and unsaturated conditions. A broader goal of this
research was to assist in the development of low-cost on-site
wastewater treatment technologies for EDC removal in devel-
oping Pacific countries.
Materials and methods
Equipment and experimental design
In this study, the main objective was to test the reduction of
EDCs within large columns containing permeable carbonate
sands. A secondary objective was to test whether estrogenic
EDC removal would be affected by the addition of other soils/
materials (e.g. basalt soil, coconut fibre) to some of the sand
columns. Basalt sediment and coconut fibre were used as
amendments as these are readily available and commonly
used with carbonate sand in wastewater treatment systems
throughout the Pacific.
Environ Sci Pollut Res

The columns were operated under a range of saturation
states, and some were amended with alluvial basalt soil and
coconut husk fibre. Different saturation states and media types
were used to increase retention times, adsorption capacity and
degree of aeration and/or availability of microbial habitat.
Basalt and carbonate sand sediments were shipped from
Rarotonga in the Cook Islands (21°14′05″S, 159°46′30″W)
and sterilized (Australian customs requirement) with gamma
irradiation (Steritech Pty Ltd). Sediments were loosely packed
into vertical flow PVC columns (diameter 150 mm, length
1.8 m). Columns were kept in a dark temperature-controlled
room at the South Lismore sewage treatment plant (STP),
Lismore, NSW, Australia. In all, there were seven carbonate
sand treatments, with each treatment containing two columns.
All columns contained 100 mm of 20-mm blue metal grav-
el above the carbonate sand to distribute the effluent evenly to
the underlying media. Treatment 1 (T1) was a fully unsaturat-
ed carbonate sand column, treatment 2 (T2) was similar to T1
except that it received a double dosing of influent and treat-
ment 3 (T3) was fully unsaturated sand with a 100-mm layer
of coconut fibre at the top of the column. Treatment 4 (T4)
contained an unsaturated homogenous mixture of sand and
basalt sediment. Treatment 5 (T5) contained unsaturated car-
bonate sand with a 100-mm basalt sediment layer 900 mm
from the top of the column. Treatment 6 (T6) contained car-
bonate sand with a saturated zone of 800 mm, created by
lifting the outlet tubing to the desired height, and treatment 7
(T7) contained carbonate sand with a 1500-mm saturated
zone. Grain size for the carbonate sand was ~81 % greater
than 0.5 mm. Basalt soil was with ~49 % of grains less than
50 μm. Chemical composition of the sediments showed that
trace metals were higher in the basalt soils. The carbonate
sands contained 12 % carbon, and the basalt had 3 % carbon;
both sediments had <0.5 % nitrogen. Grain size and soil anal-
ysis results were taken from unpublished studies by Tait et al.
(manuscript submitted for publication).
Columns were dosed regularly with effluent from the
STP’s primary sedimentation tank. Primary treated effluent
was transferred via a 25-mm pipe from the sedimentation tank
to a control valve attached to a 200-μm filter (Arkal). The
filter was intended to mimic septic tank discharge effluent
and remove larger particulate matter that had potential to clog
dosing lines. The primary treated influent was connected to a
240-L PVC barrel with a 15-mm ball float valve to maintain a
constant volume. Primary treated influent was pumped from
the barrel via a 12-V submersible pump (Amazon) to a capped
450×90 mm PVC header tank above the columns. The header
tank allowed for even distribution of effluent to columns. The
cap on the bottom of the header tank had 5 mm evenly dis-
tributed outlets inserted. The outlets were connected to 1.5 m
of 6-mm PVC tubing. The header tank would fill and feed
effluent through the tubes to distribution bars in the top
50 mm of the columns. The distribution bar consisted of a
modified 50-mL plastic syringe with a series of 10-mm holes
drilled along the bottom side. The distribution bar allowed for
less restriction on outlet tube flow and created an even dis-
persal of influent over the media. The submersible pump was
connected to a 12 V timer that was set to dose six times daily
(7:00 am, 9:00 am, 11:00 am, 3:00 pm, 6:00 pm and 9:00 pm).
The pump was set to deliver approximately 140 mL per col-
umn, which is equivalent to a hydraulic loading rate of
48 mm day
−1
. The double-dose treatment (T2) had two inlet
tubes and was given 280 mL per dose equivalent to a hydrau-
lic loading rate of 96 mm day
−1
.Figure1shows a picture of
the experimental design. Flow rates were monitored regularly
to maintain a consistent effluent flow to columns. Dosing rates
were based on the maximum daily loading rates used for car-
bonate sand sediments in the Lagoon Protection Zone for the
Cook Islands (CIMH 2009).
Initially there was an equilibration period where columns
were dosed with 2 L of primary treated effluent weekly for
6.5 weeks (20 April to 6 June 2011). The equilibration period
allowed sediments to settle within columns and a biofilm to
develop on amendments and previously sterilized sediments.
Following this period, the columns were dosed daily, as de-
scribed above, from 6 June 2011 to the end of the experiment.
The experimental assay ran for 7 weeks (1 July 2011 and
ended 21 August 2011).
Sampling and analysis
In the first 2 weeks of the experiment, a relative measure of
column residence time was obtained using a pulsed addition
Fig. 1 Experimental column design for T1 (unsaturated), T2
(unsaturated, double dosed), T3 (unsaturated, coconut), T4 (unsaturated
basalt mix), T5 (unsaturated basalt layer), T6 (partially saturated), T7
(fully saturated)
Environ Sci Pollut Res

of a conservative tracer (NaBr) added to the inlet barrel
(1680 μgL
−1
). The barrel was then disconnected from the
incoming STP effluent to prevent the dilution of the NaBr.
SamplesofBr
−
were collected daily from the inlet barrel and
every second day from the outlet of the columns.
Following the tracer addition, the collection and analysis of
influent and column outlet water for the determination of EDC
concentrations commenced. Influent samples were collected
three times a week, whilst column outlet sampling occurred
once a week. The sampling was performed for four consecu-
tive weeks, and samples were collected in 500-mm plastic
PVC bottles. After collection, samples were stored on ice
and processed within 3 h.
Processing of water samples for EDC concentrations in-
volved extraction using 4 mL (200 mg) C18 solid-phase ex-
traction (SPE) cartridges (Grace Scientific). These cartridges
were pre-conditioned by running 10 mL of acetonitrile
through the cartridge under vacuum, followed by 10 mL of
50 % v/vmethanol/milli Q water, 10 mL of methanol and then
20 mL of dilute phosphoric acid (pH<3). Extraction of EDCs
from the wastewater involved filtering 200 mL of each efflu-
ent sample through a 47-mm glass microfiber filter (GF/F
nominal pore size 0.7 μm). The filtered sample was then
passed through a conditioned SPE cartridge under vacuum.
The cartridge was then rinsed with 10 mL of the phosphoric
acid under vacuum. Cartridges were oven-dried for 2 days at
50 °C. To extract the EDCs from the dry SPE cartridges,
10 mL of 50 % acetonitrile/ethanol was passed under vacuum.
The eluate was collected in 15-mL centrifuge tubes, capped
and refrigerated.
The eluted samples were placed in a dry block heater
(Ratek) at 60 °C, evaporated down to 0.5–1mLundera
stream of dry N
2
gas and then transferred to 20-mL auto-sam-
pler vials with a glass pipette. Centrifuge tubes were washed
with 0.5–1 mL of hexane to remove any residue, and contents
were put into the sampler vials. The vials were then dried at
70 °C whilst under a stream of dry N
2
gas and derivatised with
100 μLofN,O-bis(trimethyl-silyl) trifluoroacetamide with
1 % trimethylchlorosilane (BSTFA +1 % TMCS) solution+
100 μLofpyridine(seemethodbyLietal.(2007)). Vials
were recapped with a lid containing a teflon insert and placed
in the dry block heater for 30 min at 70 °C. Vial caps were then
removed, and excess reagents were evaporated with a stream
of dry N
2
. Hexane (0.5 mL) was then added to vials, caps were
replaced and vials were shaken ready for analysis via gas
chromatography/mass spectrometry (GCMS). An automated
Agilent 7890 GCMS in EI (SIM) mode controlled by
Chemstation software measured the EDC analyte concentra-
tions. The 5975C Mass Spectrometer had a triple axis detector
installed with a 30 M J&W DB5-MS column. The carrier gas
was helium set at a constant flow rate of 1 mL min
−1
.The
temperature program starts at 80 °C, is ramped up to 180 °C at
20 °C min
−1
,isheldfor1min,thenrampedupto280°Cat
20 °C min
−1
and then held for 18 min. Samples were analysed
for estrone, estriol, estradiol and ethinylestradiol. Standards,
blanks and control samples were prepared similarly. An eight-
point calibration curve was plotted from standards, which
were injected in a descending order of 10, 8, 6, 4, 2, 1, 0.5
to 0.2 μgL
−1
. Linearity was obtained across the calibration
range for all compounds. For the E1 and EE2, the RSD of the
peak areas from 10, 8, 6, 4, 2 to 0.5 μgL
−1
were in the range of
3–15 and 3–17 %, respectively; those for E2 and E3 calibra-
tions were in the range of 3–15 % when the calibration were
from 10 to 1 μgL
−1
.
Chemical analyses were undertaken within an
ISO9001:8000 certified facility. Standards, spikes and recov-
eries were run with each batch of samples. Recoveries ranged
from 60 to 90 %, and results were adjusted for recoveries.
Analytical method was as described by Li et al. (2007).
Auxiliary measurements
Measurements were taken using handheld probes (Hach
HQ40D) in situ for pH, temperature and dissolved O
2
(DO).
Influent samples were collected three times a week, and outlet
samples were tested once a week using the same equipment.
Data analysis
One-way analysis of variance tests were performed on treat-
ment flow rate, pH and EEQ (estrogen equivalency) concen-
trations in all treatment outlets and the inlet barrel. Post hoc-
honest significant difference tests were performed when
P<0.05 or F
calc
>F
crit
to determine significant differences.
Percentage reduction of EDCs was calculated by dividing out-
let average by inlet average. All results are presented with the
standard deviation of the measurement. Estrogen equivalency
(equivalent to 17B estradiol) was defined by the sum of con-
centrations of individual estrogens (C) multiplied by their rel-
ative toxicity TEF, (EEQ = ∑(C
i
) × (TEF
i
). The TEF potency
values used were taken from Metcalfe et al. (2001). These
potency values were obtained by using yeast estrogenicity
screening assays on estrogens in sewage treatment plants to
determine TEF potency values relative to 17β-estradiol (E2).
The TEFs used were 17β-estradiol (E2)=1, 17α
ethinylestradiol (EE2)=0.38, estrone=0.14 and estriol=
0.037.
Results
Flow and geochemistry
The average flow rate from the inlet barrel to the columns was
143±19 mL dose
−1
. No significant difference (P>0.05) was
found between the flow rates to the columns. The conservative
Environ Sci Pollut Res

tracer breakthrough curves for the seven treatments are shown
in Fig. 2. The saturated treatments had the longest residence
times followed by the mixed sand basalt columns (Fig. 2).
The pH and DO of the column outlets and the inlet barrel
are shown in Fig. 3b, c. Average inlet pH and DO concentra-
tions were 7.93±0.33 and 1.27 mg L
−1
, respectively. The pH
generally increased, and the DO generally decreased with in-
creasing residence time (Fig. 3a–c). Treatment 2 (double-
dosed unsaturated carbonate sand) had the highest DO with
an average of 5.5 mg L
−1
. Treatment 7 had the lowest DO with
an average of 2.0 mg L
−1
.
Estrogens
The average EEQ of the influent barrel over the 4-week ex-
perimental period was 1.2 ng L
−1
(EEQ equivalent to 17β
estradiol). Estrone was the dominant component of the inlet
EDCs and constituted 58 % of the total EDC concentration
(Fig. 4). Estradiol was the least abundant EDC in the inlet
(6 %). Ethinylestradiol and estriol both contributed 18 % to
the total estrogenicity of the inlet sample.
The EEQs from the outlets of treatments 1–6 were lower
than EEQ for inlet water (Fig. 5). T1 had the best reduction of
EEQ with an EEQ of 0.5 ng L
−1
, followed by T5 which had an
EEQ of 0.6 ng L
−1
. Both treatments T1 and T5 were signifi-
cantly different (P<0.05) than treatment 7. In the fully satu-
rated columns (T7), there was a 6 % increase of measured
EDCs (ethinylestradiol and estradiol) to the outlet water.
Treatment (T7) had the highest EEQ of 2.4 ng L
−1
, which
was double that of the inlet wastewater which had an EEQ
of 1.2 ng L
−1
.
All treatments reduced estrone below average influent
levels. All treatments with the exception of T7 reduced estriol.
Treatments 1 (unsaturated sand)andT6(saturatedsand
800 mm) reduced all estrogens (E1, E2, EE2, E3) below av-
erage inlet influent levels. Positive fluxes of estradiol were
observed in all treatments except T1 and T6 (Fig. 4).
Discussion
Surface and groundwater contamination from EDCs has the
potential to significantly affect human and environmental
health in developing countries that are increasingly using
and disposing of EDCs. The behaviour of EDCs in carbonate
sands under saturated and unsaturated conditions is poorly
understood but is essential for developing low-tech on-site
wastewater treatment systems for the removal of EDC-laden
effluent in developing countries. The main objective of this
study was to quantify the removal of EDCs within carbonate
sand columns operated under different degrees of saturation.
A secondary objective was to test whether estrogenic EDC
removal would be affected by the addition of other soils/
materials (e.g. basalt soil, coconut fibre) to the sand columns.
Retention time was an important factor on all of the mea-
sured parameters (pH, DO and EDC concentration). In the
fully saturated columns, the DO concentration was the lowest,
and pH was similar to the inlet water. However, in the unsat-
urated carbonate sand treatments with the shortest retention
time, the pH was significantly lower, and the DO was signif-
icantly higher than the saturated columns. The decrease in pH
is likely caused by nitrification in the unsaturated sediments
through the release of H+ ions (Sharma and Ahlert 1977). The
Days after addition
0 5 10 15 20 25
Br-1 Concentrations (mg L-1) in column outlets
0
2
4
6
8
T1 - Unsaturated
T2 - Unsaturated double dose
T3 - Unsaturated coconut
T4 - Unsaturated basalt mixed
T5 - Unsaturated basalt layer
T6 - Partially saturated
T7 - Fully saturated
Fig. 2 Concentration of Br
(mg L
−1
) at the outlet of all seven
treatments following addition to a
common inlet wastewater tank.
T1 (unsaturated), T2 (unsaturated,
double dosed), T3 (unsaturated,
coconut), T4 (unsaturated basalt
mix), T5 (unsaturated basalt
layer), T6 (partially saturated), T7
(fully saturated)
Environ Sci Pollut Res