Environ. Sci. Technol. 2009, 43, 6151–6157
Occurrence and Origin of Estrogenic Isoflavones in Swiss River Waters C O R I N N E C . H O E R G E R , †,‡ FELIX E. WETTSTEIN,† ¨ HLER,‡ AND KONRAD HUNGERBU T H O M A S D . B U C H E L I * ,† Agroscope Reckenholz-Ta¨nikon, Research Station, CH-8046 Zu ¨ rich, Switzerland, and Institute for Chemical and Bioengineering, ETH Zu ¨ rich, CH-8093 Zu ¨ rich, Switzerland
Received April 9, 2009. Revised manuscript received June 11, 2009. Accepted June 29, 2009.
We report results from a systematic one-and-a-half year survey of the estrogenic isoflavones formononetin (FOR), biochanin A (BIO), daidzein (DAI), genistein (GEN), and equol in Swiss midland rivers. FOR was detected in about 90%, the other compounds in 13-56% of the weekly and fortnightly integrated flow proportional samples. Concentrations were mostly in the lower ng/L-range, with a maximum of 524 ng/L and 217 ng/L for equol and FOR, respectively. Due to dilution, concentrations were river discharge dependent, with higher numbers in smaller rivers. Total isoflavone loads were in the order of a few kg/y, and occurred mainly during summertime. A complementary river water monitoring campaign throughout the country confirmed the above findings. Circumstantial evidence points to grassland as a major emission source of FOR and BIO (the main compounds in red clover) in surface waters, e.g., their absence in wastewater treatment effluents, better correlations of their loads with grassland areas than with population equivalents, similar isoflavone ratios in river water and grassland runoff. Source apportionment was less clear for DAI, GEN, and equol. The contribution of isoflavones to the total estrogenicity of surface waters is probably small, except maybe in local rural catchments without major anthropogenic activities.
Introduction Phytoestrogens, such as isoflavones, are weakly nonsteroidal estrogenic polyphenols (1, 2) that occur naturally in a wide range of plants. Particularly legumes, such as clover (Trifolium ssp.) and soybeans (Glycine max) contain high levels of isoflavones in the g/kg concentration range. Red clover (Trifolium pratense) is a common pasture and forage crop known for its high content of formononetin (FOR) and biochanin A (BIO) (3, 4). In contrast, the main compounds in soybeans are genistein (GEN) and daidzein (DAI) (5, 6). In human and husbandry animals, FOR and BIO are demethylated to DAI and GEN, respectively. DAI is further reduced to equol. Equol is not produced in planta, but metabolized in the human gut and the rumen of domestic animals (7, 8). Table 1 compiles the structures, physicalchemical, and estrogenic properties of the most relevant isoflavones. * Corresponding author. E-mail:
[email protected]. † Agroscope Reckenholz-Ta¨nikon ART. ‡ ETH Zu ¨ rich. 10.1021/es901034u CCC: $40.75
Published on Web 07/14/2009
2009 American Chemical Society
Quantification of human (reviewed in ref 9) and husbandry animal (7) uptake of isoflavones has been extensively researched, but their environmental occurrence and exposure has not yet been investigated systematically. The presumed main pathways of isoflavones into the aqueous environment include (1) human excretion via sewer systems, (2) excretion from grazing livestock and manure application, and (3) runoff and drainage from fields cultivated with forage and grain legumes. Other point sources encompass, e.g., bleaching kraft mill effluents (10), and food factories (tofu) (11) as likely emitters of GEN and DAI. Both sources are of little importance in Switzerland. The presence of isoflavones in wastewater treatment plant (WWTP) in- and effluents was investigated by several authors (reviewed in ref 9). Several studies reported influent concentrations of GEN, DAI, and BIO that were often in the order of several hundred ng/L (12–17). Respective effluent concentrations in the above-mentioned studies were generally considerably lower, with variable removal rates of roughly 50-100%. FOR was only reported in an Italian WWTP influent at a concentration of 10 ng/L (15). The corresponding effluent samples did not contain any detectable amount of this compound, though. So far, isoflavone concentrations in manure are poorly investigated (18, 19). Burnisson et al. (18) examined several months old hog manure and found equol in the mg/L-range. Hog which were fed with 20% of red clover excreted 55% of the ingested FOR and DAI after eight hours (7). Lundh et al. (7) estimated the daily intake of isoflavones by cows to 50-100 g in Sweden, which are comparable to Swiss data (20). Overall, the excreted amounts of isoflavones are probably considerable and drastically over those of the natural steroids (0.3-1 mg/ruminant/day) (21). Assuming 25 g of isoflavones in some 50-60 L of manure excreted per cow and day, and an average manure application of 50-100 m3 ha-1 y-1, some 20-50 kg ha-1 y-1 of isoflavones may be entering an agricultural soil. This amount is comparable to the one produced in clover pastures (22), but may now be present in a much more mobile form. Hence, the risk of isoflavone input in rivers from manure application onto agricultural fields can probably not be neglected. The isoflavone emission from grass- and cropland (e.g., soybeans) by runoff and drainage water has not yet been investigated in greater detail. Some initial data was published by Erbs et al. (23), who reported isoflavone concentrations between 4 and 157 ng/L in drainage water from a red clover pasture field. In Switzerland, cattle are mainly fed with grass (both from pasture, as well as permanent and temporary grassland), and their manure put back to these areas in order to achieve closed nutrient cycles. Correspondingly, grassland areas and livestock units in a given catchment are highly correlated. It is therefore difficult to distinguish between direct isoflavone emission from plant, e.g., after cutting, or during decomposition of death plant material, and runoff from applied manure. In the following we consider grassland areas as representative of both input sources. The above-mentioned input sources result in the exposure of surface waters with isoflavones. Their occurrence, mostly in the low ng/L-concentration range, has been reported in several rivers (12–15, 24, 25). The study by Kawanishi et al. (11), with concentration of DAI and GEN up to 143 µg/L, is exceptional. In Switzerland, phytoestrogens were analyzed in the river Rhine in the 1990s (26), but did not include isoflavones. To our knowledge, no data on FOR in river water is yet available, except from our own preliminary report (23). VOL. 43, NO. 16, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6151
TABLE 1. Physical-Chemical Properties and Estrogenic Activities of the Investigated Isoflavones (-, No Data)
a Calculation with EPI-WIN. b From ref 34. c From ref 35. d From ref 36. e In analogy to ref 36. f At 4 °C unfiltered river water ((23)), own unpublished data. g From ref 1. h EC50 ) concentration of compound required to induce 50% of maximum estrogenic effect. i From ref 2.
The aim of this study was to elucidate the occurrence of isoflavones in Swiss surface waters over an extended time period of one-and-a-half years. We will present weekly and fortnightly integrated, flow proportional concentrations in rivers from the Canton of Zurich, being representative for the Swiss midlands, and complement this data with grab samples taken throughout the country. Influencing factors such as river discharge, seasonality, and water residence time will be discussed. The data are correlated with parameters indicative of the presumed major input sources, such as grassland area and population equivalent in the catchment, and specific isoflavone ratios. Finally, the ecotoxicological relevance of the presence of these compounds in surface waters will be briefly evaluated.
Materials and Methods River Water Samples. Several observation sites from the river monitoring programs of the Canton of Zurich (Office for Waste, Water, Energy, and Air, AWEL) and the Swiss government (National River Monitoring and Survey Program, NADUF) (Figure 1) were adapted to collect additional weekly (AWEL) and weekly/fortnightly (NADUF) integrated and flow proportional samples for isoflavones analysis (for further details see refs 27 and 28). The following rivers were sampled: To¨ss (at Ra¨mismu ¨ hle and Freienstein), Kempt at Winterthur, Eulach at Wu ¨ lflingen, Aabach at Mo¨nchaltdorf, Aa at Niederuster, Glatt (at Fa¨llanden, Oberglatt, and Rheinsfelden) (all located in the Canton of Zurich, and belonging to the AWEL monitoring program), Thur at Andelfingen (Canton of Zurich), Aare at Brugg, Rhine at Rekingen (both located in the Canton of Aargau), and Saane at Gu ¨ mmenen (Canton of Berne) (all part of NADUF). Water samples were gathered from April 2007 to November 2008. To assess the situation in other parts of Switzerland, specifically the (pre)alpine grasslands, grab samples were taken at 63 river sampling stations throughout Switzerland 6152
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 16, 2009
during July, 2-4, 2008 (Figure 1A). The surficial samples were taken from the middle of bridges. The sites were chosen from the hydrological data map of Switzerland (29), as to evenly cover the (pre)alpine regions. At six of these sites two grab samples were taken at different dates (Supporting Information (SI) Table S7). In addition, the five sampling sites in the river Glatt catchment (Figure 1B, red dots) were grab sampled twice (July, 11 and 22, 2008 under normal flow conditions and after a rain event, respectively). WWTP Samples. Four treatment plants were sampled in the river Glatt valley (at Fa¨llanden, Du ¨ bendorf, Kloten/ Opfikon, Niederglatt) in February, April, July, and November 2008. The WWTPs were chosen because of their close vicinity to the AWEL sampling stations. All selected plants have a Swiss standard treatment infrastructure with mechanicalbiological treatment, P-precipitation, nitrification and denitrification. Weekly, flow proportional samples were taken from the primary and secondary sedimentation basin, and the sampling periods were harmonized with those of the river water sampling sites. Analysis and Data Presentation. FOR, DAI, equol, BIO, GEN, and coumestrol were quantified as described in Erbs et al. (23). Details about the analytical method and quality control are also given in the SI. If an analyte was detectable but not quantifiable, its concentration was set equal to its limit of detection (LOD). Coumestrol was only detected in few samples (n ) 15) and is therefore not discussed any further. For AWEL and NADUF samples, isoflavone loads were calculated from the quantified concentrations and the mean weekly river water discharge, which were obtained from the sampling station protocols. For grab samples, the respectively mean daily river discharge was used (29).
Results and Discussion Isoflavone Concentrations in AWEL and NADUF Rivers. Isoflavones were detected regularly in the weekly and
FIGURE 1. (A) Map of Switzerland. The inserted box shows the region of the Canton of Zurich, which is enlarged in panel (B). River water sampling stations are visualized as follows: river Glatt catchment (AWEL, n ) 5, red dots, A-E), river To¨ss catchment (AWEL, n ) 4, green dots, F-I), and NADUF rivers (n ) 4, gray dots, J-M). Grab samples (nos. 1-63) were taken throughout Switzerland (n ) 63, black dots), as well as at stations A-E. Further details about the sampling stations are provided in Table 2 and Supporting Information Table S7, respectively. The grassland area of the total municipal area is given in percentage (Source: Statistic of the agricultural area of Switzerland for the year 2005).
TABLE 2. Concentrations of Formononetin [ng/L] for AWEL and NADUF Sampling Sites from April 2007 to November 2008
river location
no. samples analyzed/ detected
min concentration [ng/L]d
Aabach at Mo¨nchaltorf (A) Aa at Niederuster (B) Glatt at Fa¨llanden (C) Glatt at Oberglatt (D) Glatt at Rheinsfelden (E)
76/71 76/63 74/63 76/66 76/64
det det det det det
To¨ss at Ra¨mismu¨hle (F) Kempt at Winterthur (G) Eulach at Wu¨lflingen (H) To¨ss at Freienstein (I)
76/72 76/70 76/71 74/60
Thur at Andelfinden (J) Rhine at Rekingen (K) Aare at Brugg (L) Saane at Gu¨mmenen (M)c
73/72 36/30 35/33 29/29
max concentration [ng/L]d
median concentration [ng/L]d
River Glatta 132 8.4 217 5.0 26.7 det 34.4 4.2 51.9 6.0 a River To¨ss det 47.6 3.7 det 89.4 4.8 det 115 9.0 det 20.6 3.5 larger rivers in the Swiss Midlandsb det 38.9 5.1 det 44.2 det det 18.6 4.1 det 29.0 5.9
mean river discharge [m3/s]e
cumulative formononetin load [kg]f
grassland in river catchment area [km2]g
1.0 1.6 4.1 6.4 7.6
0.7 0.9 0.7 1.7 2.6
21.8 42.0 70.4 86.6 96.7
3.4 1.2 0.8 9.0
0.8 0.5 0.6 1.9
56.2 22.5 18.6 113.0
47.4 446 345 53.9
17.6 55.7 67.1 11.6
604.2 173.2 22.7 30.6
a
Monitoring network: AWEL (Office for waste, water, energy and air, Canton Zurich, Switzerland). b Monitoring network: NADUF (National River Monitoring and Survey Program). c Only samples during 2007. Letter in parentheses indicate abbreviations in Figure 1. d Out of detected samples. e April 2007 to November 2008. f After one-and-a-half years. g Source: Statistic of the agricultural area of Switzerland for the year 2005. Det: detected (0.8 < limit of detection (detected) < 2.4 ng/L).
fortnightly collected flow proportional samples of the rivers Glatt (incl. tributaries), To¨ss (incl. tributaries), Thur, Rhine, Aare, and Saane throughout the investigation period from April 2007 to November 2008. FOR was the most prominent representative in terms of occurrence-frequency (764 detects out of 853 analysis, i.e., 90%, Table 2). BIO, equol, DAI, and GEN were detected in 56, 47, 25, and 13%, respectively, of all analyses (SI Tables S3-S6). The median concentrations of FOR were comparable in these rivers and ranged from detected to 9.0 ng/L (Table 2). Whereas median concentrations of equol were similar to FOR, they were mostly below the limits of quantification (LOQ) for the other isoflavones (SI Tables S3-S6). In contrast, the maximum concentration of equol (524 ng/L; SI Table S4) was twice as high as the one for FOR (217 ng/L; Table 2). Respective numbers for DAI, BIO, and GEN were 34, 59, and 24 ng/L (SI Tables S3, S5, S6).
The mean river discharge of the investigated rivers covered a wide range from 0.8-7.6 m3/s for AWEL and from 47.4-446 m3/s for NADUF rivers, respectively (Table 2). Rivers with higher water discharge tended to exhibit lower isoflavone concentrations (often close to LOQ) than rivers with smaller discharges (Figure 2). For instance, maximum concentrations of 90-217 ng/L FOR were found in smaller rivers with median water discharges between 0.8 and 1.6 m3/s (Table 2). Additionally, concentrations were lower at times of heavy rain events. Both observations point to dilution as a main determinant of isoflavone concentrations. Isoflavone Loads in AWEL and NADUF Rivers. Similar cumulative FOR loads were obtained in Aa and Aabach (Table 2), the two tributaries of Greifensee (Figure 1B). The cumulative load at the outflow of Greifensee (Glatt at Fa¨llanden) was much smaller than what entered in the lake VOL. 43, NO. 16, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6153
FIGURE 2. Concentration of FOR [ng/L] vs weekly or daily mean river discharge [m3/s] in rivers Glatt (red), To¨ss (green), NADUF rivers (gray), and grab samples (black) in logarithmic scale. LOD ) limit of detection; LOQ ) limit of quantification. Values between LOD and LOQ were put at the level of LOD. (Table 2). This is to be expected, given the epilimnion water residence time of about 140 days (30), and an estimated FOR half-life of 41 days in natural waters at 4 °C ((23), Table 1). Calculating the remaining FOR load after degradation in the epilimnion with a two-time faster degradation rate due to a 10 °C higher average lake temperature, 0.025 kg is obtained. This result shows that the degradation in a lake must be taken into account. In contrast, the cumulative loads were additive further downstream of river Glatt, as well as in the river To¨ss and its tributaries (Table 2). At Rhine Rekingen, the cumulative FOR load was as high as 56 kg. A similar additive behavior of cumulative loads was observed for DAI and equol in river To¨ss, but not so for the other compounds in any of the investigated catchments (SI Tables S3-S6). The temporal development of (cumulative) isoflavone loads exhibited a distinct seasonal pattern. During summertime the mean FOR concentrations and the mean river discharges (and consequently also the isoflavone loads) were generally higher than during wintertime (Figure 3, although not significantly, based on a one sample t test). Eighty three percent and 81% of the total cumulative loads of rivers Glatt and To¨ss, respectively, were transported during the periods from April to September. The situation was less clear for equol, which occurred more regularly throughout the year in several of the investigated locations (SI Figures S1-S5). Isoflavone Concentrations in Other Swiss Rivers. To test to what extent the above findings are representative for whole Switzerland, a monitoring campaign in other Swiss rivers was conducted in July, 2-4. 2008. FOR was found in 80% of all grab samples (n ) 79, SI Table S7). BIO, equol, DAI, and GEN were detected in 60, 33, 9, and 8% of all cases, respectively. This frequency of occurrence succession is similar to the one described above. The median isoflavone concentrations in the grab samples ranged from detected up to 13.8 ng/L (SI Table S7) and were again comparable to that of AWEL and NADUF rivers. The daily river discharge of these sampling sites covered a broad range from 0.1-548 m3/s (SI Table S7). However, FOR was even quantifiable (3.2 ng/L) in the river with the highest discharge (Rhine at Diepoldsau, Figure 1A, no. 5). The sampling site with the highest FOR, BIO, and GEN concentration was river Ticino at Riazzino (South of Switzerland, Figure 1A, no. 31) with 242.6 ng/L, 297, and 43.8 ng/L, respectively. River Scha¨chen at Alpnach (pre-Alps, Figure 1A, no. 40) exhibited similar high concentrations of these compounds. All five investigated isoflavones were quantifiable at two sampling sites: River Veveyse at Vevey and river Sarine at Broc (Figure 1A, nos. 51 and 50, SI Table S7), both located in Western Switzerland. Overall, the data gathered over whole Switzerland fits well into the more detailed picture obtained for the rivers of the larger Zurich 6154
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 16, 2009
area (Figure 2). The isoflavone concentrations of the grab samples taken at the AWEL sampling sites (Figure 1B, SI Table S7) were in good agreement with those of their corresponding weekly integrated flow proportional samples. Comparison with Literature Data. The concentrations of isoflavones in Swiss rivers are in line with our earlier preliminary data (23) and can be further compared with those from surface waters in other countries. Ternes et al. (12) quantified equol in 3 out of 10 mostly grab samples from different German rivers, with a maximum of 30 ng/L. No other isoflavones were detected. In river Tiber, GEN, DAI, and BIO were present at concentrations between 1 and 7 ng/L from March to May 2002 (14), and between 1 and 5 ng/L between January and April 2004 (15). Again, no FOR was detected. BIO (up to 191 ng/L) dominated over GEN and DAI (hardly detected) in winter, spring and autumn in the Douro river estuary, whereas DAI (
