Trace determination of caffeine in surface water samples by liquid

chromatography–atmospheric pressure chemical ionization–mass
spectrometry (LC–APCI–MS)
Piero R. Gardinali
a,b,
*
, Xu Zhao
b
a
Southeast Environmental Research Center, Florida International University, University Park Campus, CP-313, Miami, Florida 33199, USA
b
Department of Chemistry, Florida International University, University Park Campus, CP-313, Miami, Florida 33199, USA
Received 11 February 2002; accepted 19 September 2002
Abstract
A new method based on liquid–liquid extraction (LLE) coupled to reverse phase liquid chromatography and atmospheric pressure
chemical ionization mass spectrometry (LC–APCI –MS) has been applied to determine trace amounts of caffeine (1,3,7-trimethylxanthine)
in surface water samples from a near coastal ecosystem such as Biscayne Bay, Florida. The rational behind the development of such method
will be to evaluate the use of unmetabolized caffeine as a potential dissolved phase tracer of human waste contamination. The method allows
for the determination of caffeine at levels as low as 4.0 ng/l (ppt) in both salt and freshwater by extracting and concentrating a 1-l water
sample to a final volume of 500 Al and using HPLC separation coupled to an atmospheric pressure chemical ionization mass spectrometry
(APCI –MS) system operated in selected ion monitoring (SIM) for the protonated molecular ions (M+ H
+
). Samples from different portions
of Biscayne Bay and the Miami River, one of its major tributaries, were analyzed and caffeine was detected in those areas previously
identified for consistently exceeding the water quality criteria for fecal coliform bacteria contamination. The caffeine concentration in the
samples with positive detection was generally low at levels equal or lower than 41 ng/l. However, there is a marked difference between
samples collected in open bay areas and those collected from the Miami River.
D 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Caffeine; Dissolved phase tracers; Wastewater; LC–APCI –MS
1. Introduction
Monitoring and tracking the infiltration of domestic
wastewater and untreated sewage into marine coastal eco-
systems has been a crucial challenge for environmental
managers whose efforts are devoted to protect the delicate
balance between biological systems at risk and the inevi-
table urban colonization of coastal areas. Molecular tracers
are unique chemical compounds that can be used to
unequivocally distinguish the contributions of different
sources to contaminant burdens in specific environments
(Eganhouse, 1997; Takada and Eganhouse, 1998; Standley
et al., 2000). As an example, the effect of anthropogenic
inputs from urban runoff, agricultural and industrial activ-
ities and direct human waste contamination on natural
waters in coastal ecosystems is usually evaluated by mon-
itoring traditional water quality parameters such as elevated
inorganic nutrient levels (Barrett et al., 1999), microbial
counts (Santiago-Mercado and Hazen, 1987; Lipp et al.,
2001; Paul et al., 2000), the presence of synthetic organic
chemicals (Sheldon and Hites, 1978; Eganhouse et al.,
1983; Albaiges et al., 1986; Ding et al., 1999; Standley et
al., 2000), fecal sterols (Grimalt et al., 1990; Chan et al.,
1998; Maldonado et al., 1999) and other emergent contam-
inants such as pharmaceuticals and hormones (Halling-
Sørensen et al., 1998; Hirsh et al., 1999; Stumpf et al.,
1999; Ternes et al., 1999). Although the occurrence of these
contaminants and their relative concentrations is a good
indicator of general environmental degradation, their ability
to unequivocally predict or isolate contribution of individ-
ual, more specific point sources is often hindered by their
multiplicity of origins (Barrett et al., 1999), their partition to
sedimentary phases (Vivian, 1986) or environmental stabil-
0160-4120/02/$ - see front matter D 2002 Elsevier Science Ltd. All rights reserved.
PII: S0160- 4120( 02) 00080- 6
* Corresponding author. Southeast Environmental Research Center,
Florida International University, University Park Campus, CP-313, Miami,
FL 33199, USA. Tel.: +1-305-348-6354; fax: +1-305-348-3772.
E-mail address: gardinal@fiu.edu (P.R. Gardinali).
www.elsevier.com/locate/envint
Environment International 28 (2002) 521–528
ity (Gabutti et al., 2000). Caffeine, 1,3,7-trimethylxanthine
(Fig. 1), is ranked number one drug worldwide and is
usually employed as a stimulant commonly found in coffee,
tea, cocoa, soft drinks and chocolate. It is also a component
in hundreds of prescription and over-the-counter drugs,
ranging from analgesics to cold medicines (Gilbert et al.,
1976; Bunker and Williams, 1979; Stavric et al., 1988;
Seiler et al., 1999; Weinberg and Bealer, 2001). The average
human consumes considerable amounts of caffeine. From
coffee alone, an average American consumes 131 mg of
caffeine per day. However, caffeine is extensively metabo-
lized by humans with only approximately 3% excreted
unchanged in the urine (Tang-Liu et al., 1983). Never-
theless, there is far more caffeine introduced to the sewage
system by disposal of unconsumed coffee, tea or soft drinks
down the sink, and rinsing of coffee pots and cups (Seiler et
al., 1999). This fact alone, combined with the limited
efficiency of septic systems, provides a potential way of
tracing human-derived wastewater intrusions in coastal
ecosystems by analyzing the concentrations of caffeine in
surface and ground water. Although the incidence of caf-
feine occurrence in sewage, surface, ground, and waste-
waters has been sparsely reported in the literature (Sheldon
and Hites, 1978; Buska et al., 1994; Seiler et al., 1999;
Standley et al., 2000; Ternes et al., 2001; Weigel et al.,
2001, Siegener and Chen, 2002), there is not enough
evidence pointing out the utility of caffeine as water tracer.
Inadequate detection limits and the lack of specific method-
ology combined with the fact that water treatment drasti-
cally reduces the caffeine contamination in wastewater
(Ternes et al., 2001) have prevented further interpretation
of the isolated occurrences previously reported. There is
Fig. 1. Structure of caffeine (a) and atrazine-d5 (b) and their respective full scan mass spectra obtained under the LC–APCI
+
–MS conditions described in the
text.
P.R. Gardinali, X. Zhao / Environment International 28 (2002) 521–528 522
very limited data on the fate of caffeine in the natural
environment. One of the few reports on caffeine degradation
by a strain of Pseudonomas putida isolated from sewage
showed that the parent compound is slowly degraded and
transformed to a variety of xanthines (Ogunseitan, 1996).
Since caffeine is most likely to persist in the water column
largely because of its high solubility (13.5 g/l), low octanol-
water partition coefficient (log Kow= 0.01, Gossett et al.,
1983) and negligible volatility (Weinberg and Bealer, 2001),
it fits the profile for a good, stable, dissolved marker directly
related to human activities with no potential biogenic
sources. This is also of particular importance for coastal
environments where septic tanks contribute large amounts
of the wastewater discharges in comparison with treated
municipal wastewaters (i.e. the Florida Keys). Analysis of
caffeine has been traditionally performed by liquid and gas
chromatography (Seiler et al., 1999; Bendriss et al., 2000;
Weigel et al., 2001), capillary electrophoresis (Horie and
Kohata, 2000) or liquid chromatography coupled to mass
spectrometry (Bellar and Budde, 1988; Kanazawa et al.,
2000) or to tandem mass spectrometry (Ternes et al., 2001).
The best reported UV detection limits for caffeine by HPLC
(k = 280 nm), which are in the order of 0.05 Ag/l (Patsias
and Papadopoulou-Mourkidou, 2000), are still inadequate to
determine the trace amounts of caffeine likely to be found in
environmental samples (ca. low parts per trillion). Caffeine
detection in the sub part-per-billion range has also been
reported by Seiler et al. (1999). The protocol requires the
extraction of large volumes of water to achieve detection
limits of 0.04 Ag/l using gas chromatography mass spec-
trometry (GC/MS) in the selected ion monitoring mode.
While there is a great improvement over traditional HPLC
techniques, the reported occurrences of caffeine are gener-
ally very close or below this reported detection limit, thus
making the interpretation of the results troublesome.
Based on the factors mentioned before, and the difficul-
ties associated with the available methodology for the
selective analytical detection of environmentally relevant
concentrations of caffeine, the proposed investigation will
focus on the development and validation of a simple
analytical protocol for the detection of trace levels of
caffeine in surface waters using direct liquid–liquid extrac-
tion (LLE) pre-concentration followed by selective detec-
tion using high performance liquid chromatography and
atmospheric pressure chemical ionization (APCI) mass
spectrometry (HPLC–MS).
2. Experimental
2.1. Reagents and standards
Caffeine, (1,3,7-trimethylxanthine, Fig. 1a) was pur-
chased from Fisher Scientific (Suwannee, GA, USA). Deu-
Fig. 2. Sample locations along Biscayne Bay and the Miami River, Florida.
P.R. Gardinali, X. Zhao / Environment International 28 (2002) 521–528 523
terated atrazine (atrazine-d5, Fig. 1b) used as a surrogate
standard was purchased as a certified standard solution from
Dr. Ehrenstorfer (Augsburg, Germany). Standard stock
solutions of caffeine and atrazine-d5 at concentrations of
100 ng/Al in water were prepared and stored in the dark at
< À10 jC for up to 1 year. Working standard solutions used
to prepare the calibration and fortification standards were
prepared bimonthly and routinely checked for degradation.
Anhydrous sodium sulfate, granular, combusted at 450 jC
for >6 h and sodium hydroxide were obtained from Fisher
Scientific. Methanol and methylene chloride Optimak
grade or equivalent were obtained from Fisher Scientific
and used without further purification. Double deionized,
distilled water was used for the preparation of blanks and
caffeine-fortified samples.
2.2. Sample collection
Surface water samples were collected from 14 stations
along Biscayne Bay Florida and one of its mayor freshwater
tributaries, the Miami River, between June and August 2000
(Fig. 2). Samples were specifically collected from both open
bay areas and sites potentially affected by sewage contam-
ination at the Miami River. All samples were collected in 1-l
pre-cleaned amber glass bottles from a small boat at a depth
of 30 cm below the water surface to avoid collecting the
surface microlayer. All samples were stored in the dark at
< 4 jC until analysis when they were not immediately
extracted after arriving to the laboratory.
2.3. Sample preparation
Since the purpose of the study was to evaluate the
caffeine concentration in dissolved phase, all samples were
filtered through a combusted 0.45-Am glass fiber filter
(Whatman, Clifton, NJ, USA) before analysis. Typically, a
1-l volume of the filtered surface water sample was trans-
ferred to a 2-l separatory funnel, and the appropriate amount
of internal standard (atrazine-d5) was added. After raising
the pH of the sample to 8–9 with a 0.1 M sodium hydroxide
solution, the samples were extracted using liquid–liquid
extraction with three portions of 50 ml methylene chloride.
The organic extracts were concentrated to a volume of 2–5
ml using a water bath and a Snyder column and further
evaporated to dryness using a gentle stream of purified
nitrogen. After reconstituting the samples to 500 Al of 1:1
methanol/water, they were analyzed by LC/MS as described
below.
2.4. Analytical determination
Quantitative determination of caffeine in the final
extracts was carried out by HPLC–APCI –MS in positive
mode under selected ion monitoring (SIM) using a Navi-
gator aQa system (Finnigan, Manchester, UK). The HPLC
system (Thermoquest, San Jose, CA, USA) consisted of a
P4000 model quaternary pump, an AS300 autosampler
equipped with a 20-Al loop and a UV600 PDA detector.
A Luna HPLC column (150 Â4.6 mm I.D.) packed with a
bonded C18 phase (5 Am) was used for the chromatographic
separation (Phenomenex, Torrance, CA, USA). The mobile
phase was programmed to run in a linear gradient from 30%
methanol/70% water to 100% methanol at a flow-rate of 1
ml/min. The total run time required for the analysis was 10
min with caffeine and atrazine-d5 eluting at 3.97 and 8.83
min, respectively. APCI source parameters were optimized
for the detection of caffeine as follows: cone voltage was set
at 15 V (Fig. 3), the corona needle was kept at 4.07 kVand
the source temperature was held at 300 jC. Nitrogen was
used as the nebulizing and drying gas. The selected ions
monitored and the ion ratio criteria for the compounds of
interest are shown in Table 1. To further improve the method
sensitivity, acquisition time ranges for caffeine and atrazine-
d5 were split from 0 to 6 min and from 6 to 10 min,
respectively.
Internal standard linear calibration plots were constructed
using eight points at 2.5, 5, 10, 25, 50,100, 200 and 400 pg/
Table 1
Spectral recovery and minimum detection data for the LC–APCI
+
–MS detection of caffeine in surface water samples
Analyte tR (min) Monitored ions Dwell Mean recovery (%) LOD (Ag/l)
a
Range of calibrated
M1 M2 Ratio M2/M1
time (ms) (%RSD) n = 6 concentrations (ppt)
Caffeine 3.93 195 196 0.10 100 89 (8.3) 0.004 2.5–400
Atrazine-d5 8.83 221 223 0.32 100 N/A N/A N/A
a
LOD was calculated by using a signal-to-noise ratio of 10 (the ratio between the peak intensity and the average noise level).
Fig. 3. APCI
+
source cone voltage optimization for the detection of
caffeine.
P.R. Gardinali, X. Zhao / Environment International 28 (2002) 521–528 524
ml. Since LC/MS systems are prone to background con-
tamination, a detection limit calibration solution (CS0, 2.5
pg/ml) was used to check instrument performance at low
concentrations. The APCI source was cleaned when the
signal-to-noise ratio obtained for this calibration solution
was below 3. Assessment of method performance was
conducted by analyzing saltwater and freshwater blanks
and fortified samples. To avoid the detection of false
positives due to naturally occurring organics, especially in
the river samples, the limit of detection (LOD) was calcu-
lated by using a caffeine signal-to-noise ratio of 10. Cali-
bration equations obtained in SIM mode, their appropriate
R
2
values and the overall method performance parameters
are presented in Table 2.
3. Results and discussion
There have been several reports on the determination of
low levels of caffeine in aqueous samples in particular for
surface, ground and wastewaters (Sheldon and Hites, 1978;
Buska et al., 1994; Seiler et al., 1999; Standley et al., 2000;
Weigel et al., 2001), and caffeine is often detected in surface
waters analyzed for other pharmaceuticals (Ternes et al.,
2001). Although a wide variety of methods were used and
the detection limits are not always provided, the environ-
mental concentrations of caffeine range from ‘‘traces’’
(Sheldon and Hites, 1978) to as high as 3.2 Ag/l (Buska et
al., 1994). The most complete report on caffeine occurrence
in surface waters (Standley et al., 2000) reported a range of
concentrations between < MDL and 0.115 Ag/l with a mean
of 0.013 F0.028 Ag/l. Despite the fact that the method
detection limit is not reported in the data, it is safe to
assume that concentrations as low as 0.013 Ag/l can be
detected. More recently, Ternes et al. (2001) reported con-
centrations of caffeine in rivers and wastewaters in Germany
ranging from 0.010 Ag/l (LOD in river waters) to 147 Ag/l.
So far, this is the lowest detection limit reported for caffeine
in aqueous matrices available today in the open literature.
Nevertheless, since the detection of small changes in con-
centrations is needed for the effective interpretation of
caffeine data for tracing purposes, many of the studies often
conclude that the presence of the compound is indicative of
domestic waste water contamination, but fail short to predict
the extent of it. The importance of developing a robust
method for consistent detection of caffeine at the low parts
per trillion levels is the capability to establish both the
existence of an environmental impact and the potential to
assess changes in relation to water transport.
3.1. Method development
Atmospheric pressure chemical ionization is a very
amenable technique for the detection of caffeine. Mass
spectra of both the target compound and the internal stand-
ard are extremely simple with strong abundant protonated
pseudo-molecular ions [M+ H
+
] being the major ionization
products (Fig. 1a and b). In addition, the presence of isotope
clusters provided an adequate way to perform ion ratio
confirmations (see Table 1 for ratios). Since the LC eluent
had a relatively simple composition (no modifiers were
used), both the formation of adducts and the chemical
background were kept at a minimum. Optimization of the
source parameters showed that the voltage applied to the
entrance cone rather than the ionization energy and the
source temperature mostly controls ion detection. Ion trans-
port to the mass analyzer was optimum at a cone voltage of
15 V for both caffeine and atrazine-d5. Changes in the
composition of the mobile phase were the key issue for the
optimization of the LC separation. Retention time for
caffeine and atrazine-d5 using an initial mixture of 50:50
methanol/water were 2.5 and 15.4 min, respectively. Given
the selectivity of the mass spectrometer detector, this sep-
aration was adequate, but the detection of caffeine at low
levels in the natural samples was often hindered by the
ionization of naturally occurring organics present in the final
sample extract. To circumvent this difficulty, the elution
program was changed to the optimized 30:70 methanol/
water initial mobile phase composition. With this modifica-
tion, the retention times of caffeine and atrazine-d5 changed
to 3.93 and 8.83, respectively. The increased retention for
caffeine improved the detection by an almost order of
magnitude by separating it from the solvent front (less
chemical background at the detection window). In addition,
the early elution of atrazine-d5 provided the added advant-
age of an overall shorter run time of 10 min.
3.2. Initial method performance
The initial method performance was evaluated by extract-
ing fortified fresh and saltwater blanks and fortified natural
samples. The overall results (Table 1) indicated good recov-
ery of caffeine (89%, 8.3% RSD, n = 6) from all samples
independently of matrix composition. Reproducibility was
also checked on a more limited basis by analyzing a duplicate
of one of the Miami River Samples (MR2, Table 3). Repli-
cation between duplicates was very good (31.5 F2.65,
8.33% RSD). Although it is very difficult to eliminate
sources of caffeine contamination in a laboratory, only a
few of the blanks tested during the study showed traces of
caffeine and always at concentrations below the 4 ng/l LOD.
3.3. Linearity and limit of detection
The caffeine calibration curve data obtained for six
replicate analyses is presented in Table 2. Based on the
Table 2
Internal standard-based calibration performance
Compound Equation Slope
(RSD%)
R
2
Average
RRF
RRF
(%RSD)
Caffeine y = 1.9558x + 0.0176 4.75 0.9997 1.845 5.89
P.R. Gardinali, X. Zhao / Environment International 28 (2002) 521–528 525
relative standard deviation obtained for both linear calibra-
tions (slope = 1.9558 F4.75%) and response factor-based
quantitation (RRF = 1.845 F5.89%), an excellent linearity
was obtained for the concentration range tested (5–500 pg/
Al). It is important to note, however, that the calculated LOD
corresponds to a caffeine signal-to-noise ratio of 10 (4 ng/l),
while responses of the lowest calibration solution (CS0, 2.5
pg/AL) will produce detection limits in the order of 1.25 ng/l
at a signal-to-noise ratio of >3. However, since instrument
performance at this level is less reproducible, a decision was
made to use this calibration solution as an instrument
sensitivity check rather than an extra calibration point.
Typical reconstructed mass fragmentograms for a blank,
the sensitivity check solution and a sample from the Miami
River are shown in Fig. 4.
3.4. Application to real samples
Upon completion of the method development phase, its
performance was tested by analyzing a set of surface water
samples from Biscayne Bay and the Miami River (Fig. 2).
The results, presented in Table 3 and summarized in Fig. 5,
confirm the presence of caffeine in 82% (9/11) of the
stations surveyed. Concentrations of caffeine ranged from
< LOD to 41.2 ng/l ( < 0.004 to 0.041 Ag/l) with levels in the
open bay areas being in average four times lower than the
ones observed in the river. Since the amount of data
generated in this study is limited, interpretation of spatial
trends other than the difference between the river and the
bay waters is difficult; however, the fact that caffeine
concentrations at station MR1 were consistently elevated
(41.2 and 21.8 ng/l) in two separate sampling episodes (June
Table 3
Concentrations of caffeine (1,3,7-trimethylxanthine) in surface water samples from Biscayne Bay, Florida
Sampling Site Location Classification Sampling date Caffeine
Latitude Longitude
concentration (ng/l)
Miami River 1A (MR1A) 25j46V187WN 80j11V382WW Mixed, estuarine June 23, 2000 41.2
Miami River 1B (MR1B) 25j46V187WN 80j11V382WW Mixed, estuarine August 14, 2000 21.8
Miami River 2 (MR2) 25j46V957WN 80j12V938WW Freshwater, river August 14, 2000 31.8 (2.65)
a
Miami River 3 (MR3) 24j47V189WN 80j13V434WW Freshwater, river August 14, 2000 24.1
Miami Beach Marina (MBM) 25j46V008WN 80j08V273WW Saltwater, marina June 23, 2000 7.47
Mc Arthur Causeway (MCC) 25j46V461WN 80j08V657WW Saltwater, waterway June 23, 2000 6.16
Governor’s Cut (BBGC) 25j45V867WN 80j07V998WW Saltwater, waterway June 23, 2000 <LOD
Biscayne Bay 126 (BB126) 25j40V074WN 80j12V201WW Saltwater, open bay June 23, 2000 <LOD
Biscayne Bay 129 (BB129) 25j44V400WN 80j11V160WW Saltwater, open bay June 23, 2000 11.9
Biscayne Bay 130 (BB130) 25j45V630WN 80j10V111WW Saltwater, restricted bay June 23, 2000 11.9
Biscayne Bay 131 (BB 131) 25j48V063WN 80j10V086WW Saltwater, restricted bay June 23, 2000 7.69
Red 3 Marker (BBR3) 25j45V856WN 80j08V775WW Saltwater, waterway June 23, 2000 <LOD
a
Result is the average of two determinations. The number in parenthesis is the standard deviation. < LOD indicates a detected concentration below the 4
ng/L detection limit at a S/N ratio of 10.
Fig. 4. Typical reconstructed mass chromatograms for a saltwater blank, the
lowest calibration solution (CS0) and a Miami River sample.
Fig. 5. Caffeine concentration distributions in surface water samples from
Biscayne Bay and the Miami River, Florida.
P.R. Gardinali, X. Zhao / Environment International 28 (2002) 521–528 526
and August 2000) indicates the existence of a sustained flux
of human-derived wastewater contamination along the river
water. These results are in good agreement with the long-
term history of contamination documented in the Miami
River where elevated bacterial counts are often observed in
conjunction with recurrent episodes of deteriorated water
quality (Hefty, 2000).
4. Conclusions
The analytical method developed in the present work
allows for the identification and quantitation of unmetabol-
ized caffeine in surface water samples in a simple and
sensitive way. LLE extraction of a 1-l water sample volume
and solvent exchange to a compatible mobile phase solvent
allow for the determination of caffeine at low parts per
trillion levels (4 ng/l LOD), making the interpretation of
environmentally relevant concentrations possible. The over-
all method performance based on the figures of merit
presented was very good both in terms of precision and
accuracy. The analysis of real samples from Biscayne Bay
and the Miami River revealed a fourfold difference between
areas not commonly affected by nutrient enrichment or
sewage inputs versus areas chronically influenced by sew-
age discharges and elevated eutrophication. Monitoring
such small changes may be particularly important in
urban-stressed coastal environments like Biscayne Bay or
the Florida Keys where frequent accidental ruptures of
sewage lines, pump station overflows or elevated numbers
of inefficient septic tank systems are common. Further
studies aimed to assess the co-occurrence of caffeine and
other traditional water quality parameters and to evaluate the
extent of the river influence to the adjacent bay waters are
already underway.
Acknowledgements
The authors would like to thank Florida International
University Advanced Mass Spectrometry Facility (AMSF)
for providing analytical support and the Southeast Environ-
mental Research Center (SERC) for financial support.
SERC contribution # 02-184.
References
Albaiges J, Casado F, Ventura F. Organic indicators of groundwater pollu-
tion by a sanitary landfill. Water Res 1986;20:1153–9.
Barrett MH, Hiscock KM, Pedley S, Lerner DN, Tellam JH, French MJ.
Marker species for identifying urban groundwater recharge sources: a
review and case sturdy in Nottingham, UK. Water Res 1999;33:
3083–97.
Bellar T, Budde W. Determination of nonvolatile organic compounds in
aqueous environmental samples using liquid chromatography/mass
spectrometry. Anal Chem 1988;60:2076–83.
Bendriss E, Markoglou N, Wainer IW. Liquid chromatographic method for
the simultaneous determination of caffeine and fourteen caffeine metab-
olites in urine. J Chromatogr, B 2000;746:331–8.
Bunker M, Williams M. Caffeine contents in common beverages. J Am
Diet Assoc 1979;74:28–32.
Buska P, Barber L, Schroeder M, Becker L. Organic Compounds Down-
stream from a Treated-Wastewater Discharge near Dallas, Texas, March
1987. U.S. Geological Survey, Water Resources Investigations Report
93-4194, Austin, TX, USA; 1994.
Chan KH, Lam MH, Poon KW, Yeung HY, Chiu TKT. Application of
sedimentary fecal stanols and sterols in tracing sewage pollution in
coastal waters. Water Res 1998;32:225–35.
Ding WH, Wu J, Semadeni M, Reinhard M. Occurrence and behavior of
wastewater indicators in the Santa Ana River and the underlying aqui-
fers. Chemosphere 1999;39:1781–94.
Eganhouse R. Molecular markers in environmental geochemistry. In: Egan-
house RP, editor. ACS symposium series, vol. 671. Washington, DC:
American Chemical Society; 1997. p. 1–20.
Eganhouse RP, Brumfield DL, Kaplan IR. Long-chain alkylbenzenes as
molecular tracers of domestic wastes in the marine environment. Envi-
ron Sci Technol 1983;17:523–30.
Gabutti G, De Donno A, Bagordo F, Montagna MT. Comparative survival
of faecal and human contaminants and use of staphylococcus aureus as
an effective indicator of human pollution. Mar Pollut Bull 2000;40:
697–700.
Gilbert R, Marshman J, Schweider M, Berg R. Caffeine contents of bev-
erages as consumed. J Can Med Assoc 1976;114:205–8.
Grimalt JO, Fernandez P, Bayona JM, Albaiges J. Assessment of fecal
sterols and ketones as indicators of urban sewage inputs to coastal
waters. Environ Sci Technol 1990;24:357–63.
Gossett RW, Brown DA, Young DR. Predicting the bioaccumulation of
organic compounds in marine organisms using octanol/water partition
coefficients. Mar Pollut Bull 1983;14:387–92.
Halling-Sørensen B, Nors Nielsen S, Lanzky P, Ingerslev F, Holten Lu¨tz-
hoft H, Jørgensen S. Occurrence, fate and effects of pharmaceutical
substances in the environment—a review. Chemosphere 1998;36:
357–93.
Hefty, L. Biscayne Bay Surface Water Quality Monitoring Program, Per-
sonal Communication. Miami-Dade County Environmental Resources
Management (DERM) 2000.
Hirsh R, Ternes TA, Haberer K, Kratz KH. Occurrence of antibiotics in the
aquatic environment. Sci Total Environ 1999;225:109–19.
Horie H, Kohata K. Analysis of tea components by high-performance
liquid chromatography and high-performance capillary electrophoresis.
J Chromatogr, A 2000;881:425–38.
Kanazawa H, Atsumi R, Matsushima Y, Kizu J. Determination of theophyl-
line and its metabolites in biological samples by liquid chromatography-
mass spectrometry. J Chromatogr, A 2000;870:87–96.
Lipp EK, Farrah SA, Rose JB. Assessment and impact of microbial fecal
pollution and human enteric pathogens in a coastal community. Mar
Pollut Bull 2001;42:286–93.
Maldonado C, Dachs J, Bayona JM. Trialkylamines and coprostanol as
tracers of urban pollution in waters from enclosed seas: the Mediterra-
nean and Black Sea. Environ Sci Technol 1999;33:3290–6.
Ogunseitan OA. Removal of caffeine in sewage by Pseudonomas putida:
implications for water pollution index. WJ Microbiol Biotechnol 1996;
13(3):251–6.
Patsias J, Papadopoulou-Mourkidou E. Development of an automated on-
line solid-phase extraction-high-performance liquid chromatographic
method for the analysis of aniline, phenol, caffeine and various selected
substituted aniline and phenol compounds in aqueous matrices. J Chro-
matogr, A 2000;904:171–88.
Paul JH, McLaughlin MR, Griffin DW, Lipp EK, Stokes R, Rose JB. Rapid
movement of wastewater from on-site disposal systems into surface
waters in the lower Florida Keys. Estuaries 2000;23:662–8.
Santiago-Mercado J, Hazen T. Comparison of four membrane filter meth-
ods for fecal coliform enumeration in tropical waters. Appl Environ
Microbiol 1987;53:2922–8.
P.R. Gardinali, X. Zhao / Environment International 28 (2002) 521–528 527
Seiler RL, Zaugg SD, Thomas JM, Howcroft DL. Caffeine and pharma-
ceuticals as indicators of wastewater contamination in wells. Ground
Water 1999;37:405–10.
Sheldon LS, Hites RA. Organic compounds in the Delaware River. Environ
Sci Technol 1978;12:1188–94.
Siegener R, Chen RF. Caffeine in Boston Harbor seawater. Mar Pollut Bull
2002;44:383–7.
Standley LJ, Kaplan LA, Smith D. Molecular traces of organic matter
sources to surface water resources. Environ Sci Technol 2000;34:
3124–30.
Stavric B, Klassen R, Watkinson B, Karpinski K, Stapley R, Fried P.
Variability in caffeine consumption from coffee and tea: possible sig-
nificance for epidemiological studies. Found Chem Toxicol 1988;26:
180–5.
Stumpf M, Ternes TA, Wilken RD, Rodrigues SV, Baumann W. Polar drug
residues in sewage and natural waters in the state of Rio de Janeiro,
Brazil. Sci Total Environ 1999;225:135–41.
Takada H, Eganhouse R. In: Meyers RA, editor. Encyclopedia of environ-
mental analysis and remediation. New York: Wiley; 1998. p. 2833–940.
Tang-Liu D, Williams R, Riegelman S. Disposition of caffeine and its
metabolites in man. J Pharmacol Exp Ther 1983;24:180–5.
Ternes TA, Stumpf M, Mueller J, Haberer K, Wilken RD, Servos M.
Behavior and occurrence of estrogens in municipal sewage treatment
plants—I. Investigations in Germany, Canada and Brazil. Sci Total
Environ 1999;225:81–90.
Ternes T, Bonerz M, Scmidt T. Determination of neutral pharmaceuticals in
wastewater and rivers by liquid chromatography–electrospray tandem
mass spectrometry. J Chromatogr, A 2001;938:175–85.
Vivian C. Tracers of sewage sludge in the marine environment: a review.
Sci Total Environ 1986;53:5–40.
Weigel S, Bester K, Huhnerfuss H. New method for rapid solid-phase
extraction of large-volume water samples and its application to non-
target screening of North Sea water for organic contaminants by gas
chromatography – mass spectrometry. J Chromatogr, A 2001;912:
151–61.
Weinberg B, Bealer B. The world of caffeine: the science and culture of the
world’s most popular drug. New York: Routledge; 2001. p. 213–316.
P.R. Gardinali, X. Zhao / Environment International 28 (2002) 521–528 528