Influence of poultry litter land application on the concentrations of estrogens in water and sediment within a watershed

Environmental ScienceProcesses & Impacts

PAPER

Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article OnlineView Journal | View Issue

aDepartment of Crop and Soil Sciences, Univ

E-mail: [email protected]; Fax: +1 770 412-bDepartment of Biological and Agricultural E

GA 31793, USAcCollege of Resources and Environmental

Nanjing, Jiangsu, 210095, P. R. ChinadDepartment of Crop and Soil Sciences, Univ

† Electronic supplementary informa10.1039/c3em30927d

Cite this: Environ. Sci.: ProcessesImpacts, 2013, 15, 1383

Received 29th November 2012Accepted 22nd April 2013

DOI: 10.1039/c3em30927d

rsc.li/process-impacts

This journal is ª The Royal Society of

Influence of poultry litter land application on theconcentrations of estrogens in water and sedimentwithin a watershed†

Qi Luo,a Paige Adams,b Junhe Lu,c Miguel Cabrerad and Qingguo Huang*a

This research studied the occurrence of estrogens in the Upper Satilla watershed, Georgia, USA, which was

impacted by poultry litter land application and discharge from a sewage treatment plant (STP) receiving

poultry wastes. Over 14 months, four estrogens in stream water, sediment, suspended particles, and STP

samples were quantified by LC/MS. Estrogens were consistently found in the STP influent with high

concentrations while they were below the detection limits in the majority of stream water, suspended

particles, and sediment. Estrone, 17b-estradiol, and estriol were found in 18% of stream water samples

with concentrations up to 46.4, 67.2, and 125 ng L�1, respectively. However, 17a-ethinylestradiol was

only detected in STP samples. Estrogens were found in 14% of suspended particle samples with the

median concentration being 27.5 ng g�1 for estrone, 104.5 ng g�1 for 17b-estradiol, and 93.9 ng g�1

for estriol. The estrogen concentrations in sediment were <4.95 ng g�1, indicating that sediment is not a

major sink for estrogens in this watershed. The quantitative analysis of the temporal and spatial

distribution of the estrogens suggests the occasional elevation of estrogens in the watershed above the

predicted-no-effect-concentrations to fish likely to be associated with litter disposal and rainfall events.

Environmental impact

Estrogens are emerging contaminants that can cause adverse effects to aquatic species and human health. The concentrations of four estrogens includingestrone, 17b-estradiol, estriol, and 17a-ethinylestradiol were measured in stream water, suspended particles, and sediment within a watershed that is heavilyimpacted by poultry farming. The data and associated statistical analysis provided insight into the temporal and spatial distribution of estrogens in relation torainfall and poultry litter land application.

Introduction

The presence of estrogens in the environment is a global concernbecause estrogens such as 17b-estradiol (E2), estrone (E1), estriol(E3), and 17a-ethinylestradiol (EE2) can disrupt the endocrinesystems of aquatic wildlife at extremely low levels.1 The reportedpredicted-no-effect-concentrations (PNECs) for protecting freshwater biota are 0.1 ng L�1 for EE2, 1 ng L�1 for E2, and 3 to 5 ngL�1 for E1.2 Steroidal estrogens have been detected in variousenvironmental matrices, with the concentrations in river waterranging from <0.1 to 277 ng L�1 and in river sediments from<0.05 to 22.8 ng g�1.3,4 Estrogens can sorb moderately to soil and

ersity of Georgia, Griffin, GA 30223, USA.

4734; Tel: +1 770 229-3302

ngineering, University of Georgia, Tion,

Science, Nanjing Agricultural University,

ersity of Georgia, Athens, GA 30602, USA

tion (ESI) available. See DOI:

Chemistry 2013

sediments, thus reducing their concentrations in the aqueousphase.5 The distribution of estrogens in riverine compartmentsmay strongly affect their potential to inuence the aquatic lives.Suspended particles with high surface area may also serve asvehicles to transport estrogens, particularly the ones with highconcentrations of total organic carbon.6,7

There are two major paths for estrogens to enter surfacewater: STP effluent and agricultural runoff. Estrogens oenremain in the STP effluent because they cannot be completelyremoved by conventional sewage treatment systems.8 Thecommon agricultural practice of applying livestock manures toland presents a well-described mode of contamination of waterresources at the eld level. The manure wastes containhormones and hormone degradation products. Once landapplied, these compounds can be transported to surface waterand pose a serious threat to the aquatic ecosystems and sensi-tive aquatic organisms such as sh, turtles, etc. at extremely lowconcentration levels (on the order of ng L�1).9 The moderate tohigh octanol–water partition coefficients of estrogens suggestthat a signicant fraction of estrogens may be retained ordissipated in soil aer land application, so the possibility of

Environ. Sci.: Processes Impacts, 2013, 15, 1383–1390 | 1383

http://dx.doi.org/10.1039/c3em30927d

http://pubs.rsc.org/en/journals/journal/EM

http://pubs.rsc.org/en/journals/journal/EM?issueid=EM015007

Environmental Science: Processes & Impacts Paper

Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article Online

estrogens being transported from soil to surface water is rela-tively low. However, these values were calculated based onlaboratory conditions which are different from real agriculturalpractices. In actual agricultural practices, estrogens are presentwith animal wastes. The colloids and high concentration oforganic matters in animal waste could facilitate the transport ofestrogens in water–soil systems. It has been shown that appli-cation of estrogens mixed with urine to soil enhances theleaching and persistence of estrogens in soil.10 Also the landapplication of estrogens mixed with wastewater that containedhigh concentrations of organic matter resulted in less sorptionof E2 in soil than the application of E2 mixed with pure water.11

Runoff from agricultural elds amended with litters frompoultry production systems is likely to introduce estrogens toaquatic systems. The reported E2 concentration was as high as1280 ng L�1 in agricultural runoff impacted by the poultry litterapplication at a rate of 7 Mg ha�1.9 In another study, the E2concentrations ranged from 20 to 2330 ng L�1 in the runofffrom grassland amended with poultry litter.12 However, thestudy of estrogens in a watershed with a large number of poultryproduction systems is limited. The estrogen concentrationswere found to exceed the PNECs for sh in 10 to 20% of watersamples impacted by rangeland grazing.13 Another study indi-cated that the concentration of hormone in runoff was affectedby agricultural management such as tillage and applicationrate.14

The objective of this study was to investigate the impact ofpoultry litter land application and discharge of STP on theoccurrence and transport of estrogens in the Upper Satillawatershed, Georgia, USA. We monitored four estrogens (E1,E2, E3, and EE2) at selected sites in the watershed fromFebruary, 2009 to March, 2010. Concentrations of the fourestrogens in stream water, suspended particles, sediment,and STP inuent/effluent were measured separately. Waterquality parameters including temperature, pH, dissolvedoxygen, and turbidity, chloride, solid, nitrate-nitrogen,

Table 1 Description of sampling sites in the Upper Satilla watershed15

Site no.Stream name(sampling location)

Area(km2)

Upstream/downstreamto STP

1 Rocky Creek (control site) 9.100 Upstream2 Little Hurricane (County Rd 552) 65.37 Upstream3 Upper Satilla Creek (Whitley Rd) 45.17 Upstream4 Pudding Creek (US Hwy 441) 204.4 Upstream5 Hurricane Creek (GA Hwy 32) 142.0 Upstream6 Inuent to STPa (pipeline) nab Inuent7 Pond na Downstream8 Seventeen Mile River 1

(GA Hwy 32)433.5 Upstream

9 Seventeen Mile River 2(GA Hwy 158)

462.1 Downstream

10 Satilla River 1 (GA Hwy 64) 927.1 Upstream11 Satilla River 2 (GA Hwy 158) 2306 Downstream12 Satilla River 3 (GA Hwy 1) 2922 Downstream13 Effluent from STP (Sear Rd) 14.54 Effluent

a Sewage treatment plant. b Not applicable.

1384 | Environ. Sci.: Processes Impacts, 2013, 15, 1383–1390

ammonia-nitrogen, orthophosphate, total phosphorus, andtotal nitrogen were also determined. Statistical and GISanalyses were conducted to probe the temporal and spatialdistribution of the estrogens in relation to land use and otherenvironmental conditions.

Experimental methodsStudy site

Site description. The Upper Satilla watershed is located insoutheast Georgia, USA and occupies a total area of 2922 km2.There are more than 440 poultry houses, and agricultureaccounts for about 23% of land use. The watershed receivesdischarge from a STP in Douglas. The STP has a permitteddischarge of 0.263 m3 s�1 of which about 50% is from a broilerprocessing plant and another 8% from a second chicken meatprocessing facility (personal communication).

Thirteen sampling sites were selected to represent differentland uses (Table 1, Fig. 1 and S1 in the ESI†).15 Briey, site 1 isthe control site without the impact of agricultural activities.Sites 2 and 3 are located near cotton/pasture lands. Sites 4 and 5represent areas with high numbers of poultry houses as well aslarge areas of cotton/pasture lands. Site 6 is the Douglas STPinuent and site 13 is the effluent. Site 7 is at a pond receivingthe STP effluent. Site 8 is upstream of the STP effluent whereassites 9, 11, and 12 are downstream of the STP discharge. Moredetailed descriptions of the study area and the sampling sitesare provided in the ESI.†

The poultry industry is managed in such a way that thepoultry growers usually remove caked litter from poultry housesaer two to ve ocks. It typically takes 6 weeks to raise oneock of chickens for sale. The poultry litter is primarily appliedto the surrounding pasture/cotton lands from February toMarch prior to the corn and cotton planting season at a rate of5 Mg ha�1. A previous study determined that poultry littercontained about 133 mg kg�1 natural estrogens.16 Based on the

Poultry housenumber

Cotton andpasture (km2)

Estimated estrogeninput (g)

Mean owrate (m3 s�1)

0 0 0 0.1130 2.40 159.6 0.9330 1.20 79.80 0.57250 5.80 385.7 1.9535 2.04 135.7 1.72na na na na0 na na na98 23.1 1536 3.96

98 23.1 1536 4.61

173 41.2 2740 6.77368 48.1 3199 20.3440 51.9 3451 26.80 na na 0.263

This journal is ª The Royal Society of Chemistry 2013


Fig. 1 Map of the Upper Satilla watershed study area. The black dots representpoultry houses.

Paper Environmental Science: Processes & Impacts

Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article Online

application rate of 5 Mg ha�1 and the areas of the pasture/cotton lands in the drainage area of each sampling site, weestimated the inputs of total natural estrogens for each site andthey are listed in Table 1.

Sampling. Thirteen sites were sampled monthly fromFebruary, 2009 to March, 2010, and one extra sample batch wascollected following a storm event inMarch, 2009. Approximately1 L of water was collected for estrogen analysis and another 1 Lfor water quality parameter analysis (ESI,† Experimentalsection). Top 25 cm sediment samples were obtained with asediment coring device (Rickly Hydrologic Inc., Columbus, OH)from all sites except sites 6, 7, and 13 in January, April, June, andOctober, 2009. The sediment samples were stored at �20 �Cuntil analysis. It is important to note that the broiler processingplant was closed permanently during our sampling period, atthe end of May 2009, and consequently all poultry houses in theUpper Satilla watershed were closed at the same time. Thisprovided us a unique opportunity to study the inuence ofpoultry farming by examining estrogen occurrences prior to andaer the broiler processing plant closure.

Materials and methodsMaterials

The estrogen standards (E1, E2, E3, and EE2 with purity >98%)were obtained from Sigma-Aldrich (St. Louis, MO). Acetonitrile,dichloromethane, hexanes, and methanol were of HPLC gradeand obtained from Fisher Scientic (Pittsburgh, PA). Stocksolutions of E1, E2, E3, and EE2 (100 mg L�1) were prepared inmethanol and stored at 4 �C. Standard solutions were madefrom the stock solutions by dilution with methanol. Solid phaseextraction cartridges C18 (6 mL, 500 mg) and Florisil (6 mL,1000 mg) were obtained from Restek (Bellefonte, PA).

Chemical analysis

Detailed information about chemical analyses in water, sus-pended particles, and sediment samples is described in the


Experimental section in the ESI.† Generally, the water sampleswere ltered with 1.0 mm pore size glass ber lters, and passedthrough C18 solid phase extraction, followed by a modiedcleanup method using a Florisil cartridge described in aprevious study.17 Suspended particles trapped on the lters werestored at �20 �C until analysis. The procedure for extractingestrogens from suspended particles and sediment involvedfreeze drying, ultrasonic solvent extraction, and Florisilcartridge cleanup. All samples were analyzed by LC/MS asdetailed in the ESI (Fig. S2†).18,19

Statistical analysis

All the data reported in this study have been corrected based onrecoveries. The data were analyzed in two categories. Sites 6, 7, 9and 13 were all associated with components of STP, thusreferred to as STP sites. The remaining sites, including site 9,were considered as agricultural sites because they were associ-ated with agricultural activities. Site 9 was included in bothgroups because it was immediately downstream to the STP andreceived a large amount of discharge from the STP, i.e. from site7 (Table 1 and Fig. S1†), and there are 23.1 km2 of cotton/pasture lands in the drainage area of site 9. A general linearmodel ANOVA test with signicant differences (P ¼ 0.05) wasconducted on SAS followed by a least signicant difference testto compare the estrogen concentrations among sites andmonths as well as the estrogen concentrations and the waterquality parameters before and aer the closure of poultryhouses and the broiler processing plant. The Proc CORR in SASwas used to analyze correlations among estrogen concentra-tions, rainfall, and water quality parameters.

Results and discussionAnalytical method performance

A total of 166 water samples, 166 suspended particle samples,and 33 sediment samples were analyzed from February, 2009 toMarch, 2010. The detection limits, quantitation limits, andrecoveries for water and solid samples are summarized in TableS1 (ESI†). The recoveries of the target compounds in watersamples ranged from 66% to 98% with standard deviations lessthan 8.7%. In solid samples, satisfactory recoveries (above 74%)were obtained and the standard deviations were within 7.8%.The quantitation limits were in a range of 0.63–2.9 ng L�1 forwater samples, and 0.12–0.60 ng g�1 for solid samples. Qualitycontrol information is available in the ESI.†

Occurrence of estrogens in agricultural sites

Among these samples, estrogens were measured above quanti-tation limits in 21 out of 124 water samples (Tables 2, 3 andS2†). In water samples collected from the agricultural sites, theaverage concentrations of E1, E2, and E3 were 1.85, 4.69, and5.04 ng L�1, respectively. EE2 was never detected in thesesamples. Similar estrogen concentrations in surface water havebeen reported by previous studies.3,20 The reported concentra-tions were 9 to 200 ng L�1 for E2, 27 to 112 ng L�1 for E1, 19 to51 ng L�1 for E3, and 73 to 831 ng L�1 for EE2 in the rst



Table 2 Number of observations, number of samples with at least one estrogen detected, maximum, and mean concentrations of each estrogen in water samplesfrom each site

Site

Water sample

ma

E1 E2 E3 EE2

nb Maxc Mean n Max Mean n Max Mean n Max Mean

1 11 0 nad na 1 62.0 6.16 0 na na 0 na na2 9 0 na na 1 11.2 1.75 0 na na 0 na na3 10 0 na na 2 21.9 4.65 0 na na 0 na na4 14 1 3.09 0.315 4 21.9 4.00 0 na na 0 na na5 8 0 na na 0 na na 0 na na 0 na na6 14 12 46.4 14.3 10 67.2 25.4 13 125 51.0 5 8.61 2.697 14 4 7.72 1.41 2 7.32 1.29 0 na na 0 na na8 14 2 3.71 0.889 2 15.5 2.52 2 23.0 3.36 0 na na9 13 1 8.09 0.752 0 na na 0 na na 0 na na10 15 1 3.27 0.349 3 36.5 4.32 0 na na 0 na na11 15 3 2.25 0.459 5 22.7 4.05 2 34.4 3.27 0 na na12 15 0 na na 0 na na 0 na na 0 na na13 14 6 11.8 3.09 3 19.8 3.04 1 7.73 0.942 2 5.69 1.41

a Number of observations. b Number of water samples with at least one estrogen detected. c Maximum concentration of estrogen in the watersample. d Not applicable.

Table 3 Number of observations, number of samples with at least one estrogen detected, maximum, andmean concentrations of each estrogen in suspended particlesamples from each site

Site

Suspended particles

ma

E1 E2 E3 EE2

nb Maxc Mean n Max Mean n Max Mean n Max Mean

1 11 1 555 50.4 1 19554 1778 0 na na 0 na na2 9 0 nad na 1 939 104 1 676 72.5 0 na na3 10 1 8.03 0.830 2 297 40.3 2 52.5 18.9 0 na na4 14 2 125 9.94 2 3.83 2.66 0 na na 0 na na5 8 1 26.2 3.30 0 na na 0 na na 0 na na6 14 3 15.0 2.37 6 105 17.5 2 26.4 3.15 0 na na7 14 2 280 22.2 2 182 21.5 0 na na 0 na na8 14 0 na na 2 247 18.3 1 167 12.0 0 na na9 13 1 32.7 2.54 1 200 15.5 0 na na 0 na na10 15 1 27.5 1.86 0 na na 0 na na 0 na na11 15 0 na na 1 277 18.6 0 na na 0 na na12 15 0 na na 0 na na 0 na na 0 na na13 14 1 45.3 3.27 0 na na 0 na na 0 na na

a Number of observations. b Number of suspended particle samples with at least one estrogen detected. c Maximum concentration of estrogen inthe suspended particle sample. d Not applicable.


Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article Online

national-scale study of emerging contaminants in the UnitedStates.3

Estrogens were measured in 13 out of 124 suspended particlesamples. High concentrations of estrogens were found to besorbed onto the suspended particles that were separated fromnatural water samples, with the highest concentration being555 ng g�1 for E1, 1.96 � 104 ng g�1 for E2, and 676 ng g�1 forE3. EE2 was however always below the detection limit.

Although the control site (site 1) was relatively pristine,estrogens were still detected in the water and correspondingsuspended particle sample collected aer the storm event inMarch 2009, while no estrogens were observed in any other


water and suspended particle samples from site 1. Other studieshave reported similar observations.13,21 For instance, Kolodziejand Sedlak reported that only one cattle excretion could elevatethe estrogen concentrations above PNECs to sh in a streamwith a ow rate of 0.012 m3 s�1.13 In our study, the ow rate atsite 1 was pretty low, i.e. 0.026 m3 s�1, even during the Marchstorm event. Under such a low ow rate condition, an incidentalexcretion by wildlife such as deer, hog, or aquatic species in theproximity of the sampling site could signicantly elevate theestrogen concentrations. High concentrations of E. coli andfecal coli in the same sample further supported this hypothesis(Table S3, ESI†).




Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article Online

In sediment samples, estrogens were found in 10 out of 33samples with concentrations ranging from <0.12 to 2.40 ng g�1

for E1 and <0.36 to 5.00 ng g�1 for E3, whereas E2 and EE2 werenever detected (Table S4 in the ESI†). These concentrations ofestrogens in sediment are in agreement with other reportedestrogen concentrations in river sediment which was impactedby agricultural activity.22 The absence of E2 indicates that theparent estrogen could effectively transform to metabolites inthe sediment.23 The very low estrogen concentrations detectedin the sediment imply that the sediment is not a sink forestrogen. The absence of the synthetic estrogen (EE2) in bothwater sample and sediment suggests that human activities arenot the major contributor of estrogens to the watershed.

Although estrogen concentrations were below detectionlimits during majority of the study period, estrogens wereroutinely detected at certain sites in stream water and sus-pended particle samples with concentrations above PNECs (EE2¼ 0.1 ng L�1, E2¼ 1 ng L�1, and E1¼ 3 to 5 ng L�1) (Fig. 2). Thisphenomenon is a concern because continuous exposure to thislevel of estrogens could result in adverse effects to aquaticspecies, especially, when more than one estrogens arepresent.24,25 Another concern is the high concentrations ofestrogens sorbed onto suspended particles. The release of theseestrogens back to the downstream aqueous phase having lowerestrogen concentrations might occur during the transport ofsuspended particles.

The occurrence of estrogens might cause undesirable effectsto aquatic life within the Upper Satilla watershed; however, thenet export of estrogens from this watershed to downstream isminimal. During the study period we did not measure anyestrogen in water, suspended particles, or sediment at site 12.Site 12 had the highest ow rate and was furthest away fromdirect estrogen contamination sources or agricultural lands(Table 1; Fig S1 and S2†). Estrogens might have been diluted anddegraded by microorganisms during transport. The dilutionfactor was equal to the site 12 ow rate divided by the site 11 owrate. The estimated dilution factor ranged from 1.3 to 2.6. Among4 out of 5 sampling occasions, in which estrogens were detectedat site 11, dilution alone could reduce estrogens to near or below

Fig. 2 Detection frequency of estrogens in water and suspended particlesamples. The estrogen was considered detected once when E1, E2, E3, EE2, or anycombination of these four estrogens was detected in the sample.


quantitation limits when estrogens transported from site 11 tosite 12. Degradation by microorganisms during transport mightalso lower the estrogen concentrations. One study reported thatmicroorganisms could degrade E2 in river water with the shortesthalf-life of 0.2 day when incubated at 20 �C.23

Impact of land use on estrogen concentrations at agriculturalsites

The means by site of each estrogen equals to the sum of thatestrogen concentration of each site divided by month. Theaveraging of each estrogen concentration at all agricultural sitesfor each month is referred to as means by time. There were nostatistically signicant differences (P > 0.05) among the meansby site regardless of sample matrix, while signicant differences(P < 0.05) were found among the means by time for E1 and E2 inwater samples (Table S5, ESI†). The fact that no signicantdifferences were found among means by site suggests thatdifferent land uses do not make a signicant difference inestrogen concentrations at the sites studied. However, severalfactors need to be taken into account when interpreting theabove results from statistical analysis. First, the means by sitewere calculated by averaging estrogen concentrations over thenumber of sampling times. However, time-dependent factors(i.e., rainfall events, closure of poultry houses) signicantlyimpact the estrogen concentrations in aqueous samples. Thelarge variance associated with time could have obscured thevariance, if any, with sites. Second, although there are somedifferences in land use between different sites, it is indeed notclearly dened. For example, cotton/pasture lands are scatteredover the most area, although numbers and density are different.So all the sampling sites except site 1 have the potential of beingimpacted by land application of poultry litter.

Variation of estrogen concentrations in agricultural sites withtime

The 14-month study period gave us an opportunity to monitorthe variation of estrogens in the environment over season.Statistically signicant differences of E1 and E2 concentrationsin natural water samples were detected among the months (P <0.05) (Table S5†). Occurrences of high estrogen concentrationscoincided with heavy rainfall (Fig. 3). When heavy rainfall wasexperienced in March, May, June, and August 2009, elevation ofestrogen concentrations in the watershed was obvious. Thedetection of estrogens at sites 2, 3, and 10 during the 2009March storm event was not surprising because these sites wereclose to the large areas of cotton/pasture land receiving litterland application. There was a 2.4 km2 cotton/pasture land in thedrainage area of site 2 and 1.2 km2 cotton/pasture land in thedrainage area of site 3. The estimated input of total naturalestrogen was 153, 79.8, and 2734 g for sites 2, 3, and 10respectively (Table 1). Farmers usually apply litters to theselands during March. The heavy rainfall in March 2009, not longaer poultry litter land application, might have washed theestrogens from the cotton/pasture lands into the streams.

Another factor that could contribute to the time-dependentvariation in estrogen concentrations is the closure of poultry



Fig. 3 Estrogen concentrations summed for all agricultural sites on each samplingevent (a) and cumulative rainfall over 0 day, 1 day, 3 days, and 5 days prior to thesampling date. After (b) MarchS09 refers to the March storm event in 2009.


Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article Online

houses. All poultry houses in the Upper Satilla watershed wereclosed at the end of May, 2009 due to the closure of the broilerprocessing plant. To address the impact of this closure, wegrouped February 2009 to August 2009 as a pre-closure time andfrom September, 2009 to March, 2010 as a post-closure time.The borderline was chosen in August because the closureoccurred inMay and it has been reported that impacts of animalwaste land application could last up to 3 months.26 The meansof the sum of each estrogen during pre-closure and post-closuretimes are presented in Fig. S3† with 95% condence intervals. Itis evident that the concentrations of E1 and E2 are signicantlydifferent before and aer the closure event (P < 0.05, Table S6,ESI†). This indicates that poultry farming and/or land applica-tion can inuence the estrogen concentrations at a signicantlevel.

Our data collected from May to August, 2009 represent aworst case scenario because all the poultry houses were emptiedaround that point of time. Our survey indicated that mostfarmers applied all the poultry litter resulting from the cleanoutof the last ocks to the surrounding agricultural land just priorto closure of the processing plant. The one time release of largeamounts of poultry litter into agricultural land would increasethe possibility of estrogens entering surface water with agri-cultural runoff as estrogen concentrations in agricultural runoffincreased with poultry litter application rate.9 In August, 2009,estrogens were still detected above the PNECs in 5 out of 10


stream water samples. In contrast, almost no estrogen wasdetected in any stream water samples aer November, 2009,which was six months aer the closure event.

Water quality parameters

In an attempt to nd a correlation between estrogen concentra-tions and water quality parameters, we measured temperature,turbidity, dissolved oxygen, chloride, orthophosphate, nitrate,ammonia, total phosphate, and total nitrogen in water samples(Table S3; estrogen concentrations and water quality parameter,ESI†). No strong correlation was found between any water qualityparameter and estrogen concentrations. The statistical analysesshow that ammonia and total nitrogen in the whole watershedwere signicantly lower aer the poultry houses were closed (P <0.05) (Table S7, ESI†). Specically, the reduction of ammoniaconcentration aer the closure of poultry houses was found atsites 3 and 9. At sites 9 and 10, total nitrogen concentration wassignicantly lower aer the closure (estrogen concentrations andwater quality parameter, Table S7, ESI†).

Fate of estrogens at STP sites

Estrogens were detected in all STP inuent (site 6) aqueoussamples and 46% of suspended particle samples (Fig. 1). Site 6is the only site which is signicantly different from the othersites for aqueous samples (P < 0.05). The concentrations were<46.4, <67.2, <125, and <8.61 ng L�1 for E1, E2, E3, and EE2respectively (Table S2†). These concentration levels were in thesame range as what had been reported in other studies.27,28 Forexample, the E1 concentration in raw sewage inuent rangedfrom <0.1 to 54.8 ng L�1, and the E2 concentration ranged from<0.1 to 22.0 ng L�1. In the STP inuent, we observed the E2concentrations to be higher than those of E1. Several studiesreported a similar trend.29,30 In one study, the reported meanconcentration of E2 was 82 ng L�1, while for E1 it was 29 ng L�1

in a wastewater treatment plant inuent. Transformation of E2and E1 to E3 could have changed their ratios. In our study, E3might be the major metabolite of E2 rather than E1 since theconcentration of E3 was generally much higher than thatof E1.31–33

The impact of the closure of the broiler processing plant onestrogen concentrations in STP inuent was analyzed by con-ducting ANOVA on the means of each estrogen concentrationaveraged by the number of sampling times for pre-closure time(February 2009 to May 2009) vs. that for post-closure time (June2009 to March 2010) (Table S6, ESI†). The result does not show asignicant difference (P > 0.05) either in aqueous phase orsuspended particles, whichmay suggest that the contribution ofthe broiler processing plant to the large amounts of estrogens inthe STP inuent is limited. As expected, EE2 was only detectedin STP inuent samples that added support to our hypothesisthat the estrogens in these samples may be primarily of humanorigin.

Estrogens were also measured in STP effluent (site 13),effluent receiving pond (site 7), and the immediate downstreamof STP (site 9). The concentrations of E1, E2, and E3 in STPeffluent and effluent receiving pond were signicantly lower




Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article Online

than in STP inuent, while EE2 was not effectively removed(Fig. S4†). It is noticeable that estrogen concentrations in thepond samples were higher than in the STP effluent at someevents (Table S3, ESI†). This might have been caused by thebirds frequently present around the pond.

Phase distribution of estrogens

No strong relationships among the concentrations of sus-pended particles, estrogen concentrations, and rainfallamounts were found by correlation analyses (Table S2, ESI†).For those samples with estrogens being detected in bothaqueous and solid phases, the sorption coefficient Kd was esti-mated by dividing the concentrations of estrogens sorbed ontosuspended particles (ng kg�1) by the dissolved concentration inaqueous phase (ng L�1). The average log Kd values are 3.55� 0.5for E1, 3.72 � 1.0 for E2, and 3.34 � 0.1 for E3. No log Kd

information for EE2 was determined since EE2 was not detectedin suspended particles. These values are higher than what werereported for the river suspended particles while in good agree-ment with the ones measured in STP suspended sludge (TableS8†).27,34–36 The variety of hydrologic conditions, characteristicsof suspended particles, and stream water states may cause theheterogeneity of log Kd in suspended particles. Moreover, thesorption of estrogens to the particles was also impacted bythe total organic carbon, salinity of aqueous phase, and theinitial concentrations of estrogens in the aqueous-suspendedparticle system.37,38

Conclusions

The concentrations of estrogens were above the PNECs for shin some samples in the Upper Satilla watershed, albeit belowthe detection limits in most samples during the 14-month studyperiod. The closure of poultry houses during the study periodrepresents a worst case scenario since the farmers might haveapplied large amounts of poultry litter to the crop land at once.Signicant differences were found in estrogen concentrationsbefore and aer the closure of the broiler processing plant andpoultry houses. The decrease of detection frequency andreduction of estrogen concentrations aer the closure indicatethat the watershed can effectively recover from the impact of thepoultry industry. These facts suggest poultry litter land appli-cation in a watershed with a dense population of poultry housescould serve as a nonpoint source to introduce estrogens intosurface water. However, no statistically signicant differenceamong the sites implies that difference of land use alone couldnot explain the occurrence of estrogens in surface water. Theassociation of rainfall and detection of estrogens in surfacewater indicates that heavy rainfall could wash the estrogensfrom agricultural land and further carry them into surfacewater. Such pulses of high estrogen concentrations may causeacute toxicity to vulnerable aquatic species, which represents aconcern from a management point of view. The high concen-trations of estrogens sorbed onto the suspended particles raisedthe concern that the sorbed estrogens may serve as a secondarysource of contamination by dissipating estrogens back to the


aqueous phase and facilitate the bioavailability of estrogens toaquatic organisms. Furthermore, the estrogens frequentlydetected in the STP effluent and pond suggest that additionaltreatment measures may be helpful to lower the release of theseemerging contaminants from wastewater treatment facilities.

Acknowledgements

This study was supported primarily by U.S. EPA Region 4 and inpart by U.S. EPA STAR grant G6M10518 and HATCH funds. Wethank Drs Ian Flitcro, Paul Raymer, and Jerry Davis from theUniversity of Georgia, Griffin for their help with data analysis.Debbie Coker, Rama Ghimire, and Michelle Gerlosky areacknowledged for helping with the laboratory work. We alsothank the investigators, Erin Lipp, Michael Jenkins, RichardLowrance, and Ethell Vereen, of the USDA-CSREES-NRI 2006-35102-17328 project for collaborating with our study.

Notes and references

1 J. P. Sumpter and A. C. Johnson, Environ. Sci. Technol., 2005,39, 4321–4332.

2 W. F. Young, P. Whitehouse, I. Johnson and N. Sorokin,Proposed Predicted-No-Effect-Concentrations (PNECs) forNatural and Synthetic Steroid Oestrogens in Surface Waters,R&D Technical Report P2–T04/1 No: EA5098 forEnvironmental Agency R&D Dissemination Centre,Swindon, UK, 2004.

3 D. W. Kolpin, E. T. Furlong, M. T. Meyer, E. M. Thurman,S. D. Zaugg, L. B. Barber and H. T. Buxton, Environ. Sci.Technol., 2002, 36, 1202–1211.

4 M. Petrovic, E. Eljarrat, M. J. Lopez de Alda and D. Barcelo,Anal. Bioanal. Chem., 2004, 378, 549–562.

5 B. S. Das, L. S. Lee, P. S. C. Rao and R. P. Hultgren, Environ.Sci. Technol., 2004, 38, 1460–1470.

6 S. Zhao, P. Zhang, M. E. Melcer and J. F. Molina,Chemosphere, 2010, 79, 420–425.

7 B. J. Robinson, J. P. M. Hui, E. C. Soo and J. Hellou, Environ.Toxicol. Chem., 2009, 28, 18–25.

8 T. Hashimoto and T. Murakami, Water Res., 2009, 43, 573–582.

9 D. J. Nichols, T. C. Daniel, P. A. Moore, Jr, D. R. Edwards andD. H. Pote, J. Environ. Qual., 1997, 26, 1002–1006.

10 S. D. Lucas and D. L. Jones, Soil Biol. Biochem., 2009, 41, 236–242.

11 B. Stumpe and B. Marschner, Sci. Total Environ., 2007, 374,282–291.

12 O. Finlay-Moore, P. G. Hartel and M. L. Cabrera, J. Environ.Qual., 2000, 29, 1604–1611.

13 E. P. Kolodziej and D. L. Sedlak, Environ. Sci. Technol., 2007,41, 3514–3520.

14 M. B. Jenkins, D. M. Endale, H. H. Schomberg andR. R. Sharpe, Sci. Total Environ., 2006, 358, 164–177.

15 E. Vereen, Jr, R. R. Lowrance, D. J. Cole and E. K. Lipp, Appl.Environ. Microbiol., 2007, 73, 1395–1403.

16 T. A. Hanselman, D. A. Graetz and A. C. Wilkie, Environ. Sci.Technol., 2003, 37, 5471–5478.




Publ

ishe

d on

22

Apr

il 20

13. D

ownl

oade

d by

Tem

ple

Uni

vers

ity o

n 24

/10/

2014

22:

55:2

9.

View Article Online

17 H. Chang, Y. Wan and J. Hu, Environ. Sci. Technol., 2009, 43,7691–7698.

18 M. S. Dıaz-Cruz, M. J. L. d. Alda, R. Lopez and D. Barcelo,J. Mass Spectrom., 2003, 38, 917–923.

19 J. Hu, H. Zhang and H. Chang, J. Chromatogr., A, 2005, 1070,221–224.

20 M. J. Benotti, R. A. Trenholm, B. J. Vanderford, J. C. Holady,B. D. Stanford and S. A. Snyder, Environ. Sci. Technol., 2009,43, 597–603.

21 L. K. Irwin, S. Gray and E. Oberdoster, Aquat. Toxicol., 2001,55, 49–60.

22 P. Labadie and E. M. Hill, J. Chromatogr., A, 2007, 1141, 174–181.

23 M. D. Jurgens, K. I. E. Holthaus, A. C. Johnson, J. J. L. Smith,M. Hetheridge and R. J. Williams, Environ. Toxicol. Chem.,2002, 21, 480–488.

24 L. J. Mills and C. Chichester, Sci. Total Environ., 2005, 343, 1–34.

25 J. V. Brian, C. A. Harris, M. Scholze, A. Kortenkamp, P. Booy,M. Lamoree, G. Pojana, N. Jonkers, A. Marcomini andJ. P. Sumpter, Environ. Sci. Technol., 2007, 41, 337–344.

26 J. Kjǽr, P. Olsen, K. Bach, H. C. Barlebo, F. Ingerslev,M. Hansen and B. H. Sørensen, Environ. Sci. Technol.,2007, 41, 3911–3917.

27 O. Braga, G. A. Smythe, A. I. Schafer and A. J. Feitz, Environ.Sci. Technol., 2005, 39, 3351–3358.


28 H. C. Chen, H. W. Kuo and W. H. Ding, Chemosphere, 2009,74, 508–514.

29 M. Clara, N. Kreuzinger, B. Strenn, O. Gans and H. Kroiss,Water Res., 2005, 39, 97–106.

30 L. Viglino, K. Aboulfadl, M. Preost and S. Sauv, Talanta, 2008,76, 1088–1096.

31 C. Baronti, R. Curini, G. D'Ascenzo, A. Di Corcia, A. Gentiliand R. Samperi, Environ. Sci. Technol., 2000, 34, 5059–5066.

32 G. D'Ascenzo, A. Di Corcia, A. Gentili, R. Mancini,R. Mastropasqua, M. Nazzari and R. Samperi, Sci. TotalEnviron., 2003, 302, 199–209.

33 A. C. Johnson, A. Belfroid and A. Di Corcia, Sci. TotalEnviron., 2000, 256, 163–173.

34 M. Carballa, G. Fink, F. Omil, J. M. Lema and T. Ternes,Water Res., 2008, 42, 287–295.

35 Y. K. K. Koh, T. Y. Chiu, A. R. Boobis, M. D. Scrimshaw,J. P. Bagnall, A. Soares, S. Pollard, E. Cartmell andJ. N. Lester, Environ. Sci. Technol., 2009, 43, 6646–6654.

36 K. I. E. Holthaus, A. C. Johnson, M. D. Jurgens,R. J. Williams, J. J. L. Smith and J. E. Carter, Environ.Toxicol. Chem., 2002, 21, 2526–2535.

37 K. M. Lai, K. L. Johnson, M. D. Scrimshaw and J. N. Lester,Environ. Sci. Technol., 2000, 34, 3890–3894.

38 G. G. Ying, R. S. Kookana and P. Dillon,Water Res., 2003, 37,3785–3791.



Documents

Influence of poultry litter land application on the concentrations of estrogens in water and sediment within a watershed