Ecotoxicology and Environmental Safety 132 (2016) 31–39

Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety

http://d0147-65

n CorrE-m

journal homepage: www.elsevier.com/locate/ecoenv

Effects of erythromycin, trimethoprim and clindamycin on attachedmicrobial communities from an effluent dominated prairie stream

M.J. Waiser n, G.D.W. Swerhone, J. Roy, V. Tumber, J.R. LawrenceEnvironment Canada, 11 Innovation Blvd., Saskatoon, SK S7N 3H5, Canada

a r t i c l e i n f o

Article history:Received 22 October 2015Received in revised form25 May 2016Accepted 26 May 2016

Keywords:AntibioticsPharmaceuticalsWaste waterRiversMicrobial communitiesRotating annular reactorsBiofilm

x.doi.org/10.1016/j.ecoenv.2016.05.02613/Crown Copyright & 2016 Published by Else

esponding author.ail address: mjwaiser@shaw.ca (M.J. Waiser).

a b s t r a c t

In this study, differing metrics were utilized to measure effects of erythromycin (ER), trimethoprim (TR)and clindamycin (CL) on the structure and function of attached Wascana Creek, SK microbial commu-nities. All three test antibiotics, especially ER, affected community structure and function of biofilmsgrown in rotating annular reactors. Biofilm thickness, bacterial biomass, and lectin binding biovolume(exopolymeric substances) were consistently less in ER treated biofilms when compared to the control.As well negative effects on protozoan numbers, and carbon utilization were detected. Finally, PCA ana-lyses of DGGE results indicated that bacterial community diversity in ER exposed biofilms was alwaysdifferent from the control. ER exhibited toxic effects even at lower concentrations. Observations on TRand CL exposed biofilms indicated that bacterial biomass, lectin binding biovolume and carbon utiliza-tion were negatively affected as well. In terms of bacterial community diversity, however, CL exposedbiofilms tended to group with the control while TR grouped with nutrient additions suggesting bothnutritive and toxic effects. This study results represent an important step in understanding antibioticeffects, especially ER, on aquatic microbial communities. And because ER is so ubiquitous in receivingwater bodies worldwide, the Wascana study results suggest the possibility of ecosystem disturbanceelsewhere.Capsule abstract: Erythromycin (ER) is ubiquitous in waterbodies receiving sewage effluent. Structureand function of microbial communities from an effluent dominated stream were negatively affected byER, at realistic concentrations.

Crown Copyright & 2016 Published by Elsevier Inc. All rights reserved.

1. Introduction

In Canada, systemic anti-infectives are among the top pre-scribed medications (Morgan et. al., 2005) while in Europe, 13,500tons of antibiotics are used each year with 65% for humans and theremaining 35% for animals (Christensen et al., 2006). Despitevariation in excretion rates and types of metabolites, some anti-biotics are excreted from humans and animals relatively un-changed. Further, many human antibiotics are not fully removedduring sewage treatment. Consequently, surface waters worldwidereceiving treated sewage effluent and animal waste contain anti-biotics in the ng/L to mg/L range. Of these antibiotics, erythromycin(ER) and trimethoprim (TR), appear to be ubiquitous in receivingwater bodies. A recent British investigation, for example, revealedaverage ER and TR concentrations of 159 ng/l and 12 ng/l, re-spectively, 1 km downstream of five sewage treatment plants(STPs) (Ashton et al., 2004). Maximum erythromycin-H2O

vier Inc. All rights reserved.

concentration (metabolite) in a United States survey of 139streams was 1.7 mg/l (Kolpin et al., 2002) while in Italy, median ERconcentrations in the Lambro and Po Rivers were 4.5 and 3.2 ng/l,respectively (Zuccato et al., 2006). Because they are so commonlyfound in receiving water bodies, many scientific reviews rank ci-profloxacin, TR, ER and sulfamethazine of particular concern(Johnson et al., 2015).

The presence of antibiotics in aquatic ecosystems is of concernfor a number of reasons. Most antibiotics interact with a biologicaltarget (membranes, enzymes) shared by humans, animals, plants(e.g. plastids) and bacteria (Brain et al., 2008). They act at relativelylow concentrations, need time to achieve effects and resist bio-degradation (Jones et al., 2004). Proliferation of antibiotic re-sistance has been linked to chronic bacterial exposure (Eggenet al., 2004; Fent et al., 2006; Gullberg et al., 2011) and high nu-trient conditions typifying water bodies receiving treated effluentmay exacerbate such proliferation (Castiglioni et al., 2008). Lowantibiotic concentrations may also stimulate or depress bacterialgene expression at the transcriptional level at concentrations sig-nificantly lower than the minimum inhibitory concentration (MIC)

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–3932

(Goh et al., 2002). Finally, their constant addition to aquatic eco-systems at rates exceeding transformation means not only thatthey are considered pseudo-persistent (Daughton and Ternes,1999) but that aquatic organisms are chronically exposed.

Not a great deal, however, is known regarding such chroniceffects as most research has been directed towards single speciestoxicity with acute exposures. Although these controlled studieshave figured largely in risk assessment, their results lack en-vironmental realism and extrapolation to communities at theecosystem level is difficult. Community-based effects assessments,however, offer the potential of evaluating ecosystem level effectsat relevant contaminant concentrations. Ideal candidates forcommunity level effects studies are microorganisms. Within riv-ers, streams and other aquatic habitats they not only serve as animportant food source for benthic invertebrates and protozoans,they are also major players in biogeochemical nutrient cycling andorganic matter biodegradation. Due to their short life cycles, toxicinduced community succession (TIS) can be observed in the la-boratory over relatively brief time scales (Backhaus et al., 2011).Microbial communities, therefore, are potentially excellent in-dicators of changes in ecosystem health.

The purpose of this study was to investigate effects of threeantibiotics, ER, TR and clindamycin (CL) on the structure, devel-opment, functioning and biodiversity of biofilm communities froman effluent dominated creek (Wascana Creek) in southern Sas-katchewan, Canada using a laboratory approach. Previous seasonalmonitoring studies here indicated that ER, CL, and TR were allpresent in ng/L and sometimes mg/L concentrations. With regard totheir toxicity, single species research has revealed that a dosage of1 mg/L ER inhibits Synechocystis sp. and Lemna minor growth by70% and 20% (Pomati et al., 2004). Other studies indicated that themedian inhibition concentration (IC50) of TR to Pseudokirchneriellasubcapitata was 1000 mg/L (Yang et al., 2008). According to hazardquotients (HQ) generated from earlier Wascana Creek research, ERand TR were present at concentrations which could present a riskto aquatic organisms (Waiser et al., 2011b). Although CL did notgenerate a hazard quotient indicative of risk, it was added to thisstudy due to its presence on all sampling dates and its persistencedownstream of the STP. It was hoped that such community basedresearch would play an important role not only in establishingeffects of these antibiotics but in structuring future environmentalrisk assessments for Wascana Creek and perhaps other similarlyaffected aquatic ecosystems (c.f. Backhaus et al., 2011).

2. Materials and methods

2.1. Study area

Wascana Creek receives tertiary treated sewage effluent fromthe city of Regina, SK sewage treatment plant. From late Octoberthrough to March most flow in the creek is treated sewage ef-fluent. Wascana Creek is considered hypereutrophic with averageseasonal total phosphorus (TP) concentration 0.24 mg/L, solublereactive phosphorus (SRP) 0.09 mg/L, ammonia-N (NH3) 0.04 mg/Land total dissolved nitrogen (TDN) 0.79 mg/L (Waiser et al., 2011a).

2.1.1. Attached microbial communities; rotating annular reactorexperiments

Biofilms recruited from Wascana Creek (algae, bacteria andexopolysaccharide) were grown for 8 weeks under controlledconditions in rotating annular biofilm reactors as described byLawrence et al. (2004). Wascana creek water was collected inAugust 2006 and November 2006 at a site approximately 30 kmupstream of the sewage treatment plant. Previous monitoringstudies indicated that none of the study antibiotics were present

here (Waiser et al., 2011b). Creek waster served as the microbialinocula as well as a source of carbon (C), nitrogen (N) and phos-phorus (P).

2.2. Experimental design

The August biofilm experiment (BEXP1) tested the effects of thethree antibiotics, each at 4 mg/L (nominal concentrations), on creekbiofilms. Control reactors with no additions were run at the sametime. The November experiment (BEXP2) repeated these threetreatments but also added 1, 2, 3 and 4 mg/L of ER (nominal con-centrations) to the investigation. Each treatment and control hadthree identical replicate reactors randomly assigned to it in ablock. Antibiotics were continuously added to the reactors usingperistaltic pumps. Nutrient controls were also run where themolar equivalent of the antibiotic treatment was added as carbon(glucose) and nitrogen (ammonium chloride) (cf Lawrence et al.,2012). These treatments are designed to test whether the impactof the contaminant is more similar to degradation and utilizationor toxicity. These nutrient controls are designated as CL control, ERcontrol and TR control respectively.

The central rotating shaft of each bioreactor contained 12identical polycarbonate slides on which the biofilms grew. At theend of the experiment, slides were randomly selected, removedand in some cases small subsections cut so that biofilms on thesesections (coupons �1 cm2) could be subjected to microscopicexamination and other assays.

2.3. Algal biomass (Chl a)

Biomass on each 1�10 cm coupon (one from each replicate)was scraped into a 50-mL centrifuge tube containing 10 mL of 90%ethanol. Chl a was extracted coupons in 90% boiling ethanol andanalysed fluorometrically using a Turner Designs Model 10-AUdigital fluorometer (Turner Designs, Sunnyvale, CA) (Waiser andRobarts, 1997).

2.4. Bacterial production (BP)

One coupon from each replicate was used to estimate BP. Eachcoupon was placed in a plastic Petri dish containing 10 mL of0.2 mm filter sterilized creek water. Approximately 30 nM of 3Hlabeled thymidine (TdR –Perkin Elmer, Waltham, MA) was addedto each dish and dishes swirled. One coupon from each treatmentwas placed into a corresponding ‘killed’ control dish to which500 mL of formalin had been previously added. Samples were in-cubated for thirty minutes, at the end of which time live sampleswere killed with 500 mL of formalin. Bacterial DNA was extractedand bacterial production was estimated using [methyl]-3H thy-midine (TdR) incorporation into bacterial DNA according to es-tablished methods (Robarts and Wicks, 1989).

2.5. Confocal laser scanning microscopy and image analysis

Coupons with biofilms attached were mounted in small petridishes using Dow Corning #3140 acid-free silicone coating (WPI,Inc., Sarasota, Fla.). Bacteria were then stained with a fluorescentnucleic acid stain (SYTO 9) (excitation wavelength, 488 nm [ex488]; emission wavelength, 522–532 nm [em 522/32]). Biofilmswere observed using a Bio-Rad MRC 1024 confocal laser scanningmicroscope (Zeiss, Jena, Germany) attached to a Microphot SAmicroscope (Nikon, Tokyo, Japan). Two water submersible lenses, a63�0.9 numerical aperture (Zeiss, Jena, Germany) and a 40� 0.55numerical aperture (Nikon) were used for biofilm examination.Laser scanning microscopy imaging was done at five positions onfive randomly chosen transects traversing each biofilm coupon.

Table 1.Oligonucleotide primers used for PCR amplification of bacterial genes.

Target gene Primer consensussequencea Position(length)

Referenceorganism

PCR fragmentsize (bp)

Targetbacteria

16S rRNAForwardb,c 5′- CCT ACG GGA

GGC AGC AG -3′Escherichiacolid

586 Bacteria

Position 341–357 (17 nt)Reverse,e,c 5′- CCG TCA ATT

CMT TTG AGT TT-3′

Position 907–927 (20 nt)

a Forward and Reverse indicate the orientation of the primers in relation to therRNA sequence.

b Preceded by a GC clamp on 5′ end for DGGE (not for sequencing)¼CGC CCGCCG CGC CCC GCG CCC GGC CCG CCG CCC CCG CCC G (40 nt) Muyzer et.al, 1993.

c Muyzer et al. (1993).d E. coli numbering according to Brosius et al. (1981).e Casamayor et al. (2000).

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–39 33

These scans were subsequently utilized to estimate biofilm thick-ness and biomass. For biomass estimates, cyanobacteria and algaewere differentiated based on wavelengths of their autofluorescentpigments and bacteria on the basis of the SYTO 9 emission wa-velength (three channel procedure - Neu et al., 2004; Lawrenceet al., 2004, 2005). Digital image analysis of resulting optical thinbiofilm sections in each channel was then performed (Neu et al.,2001).

2.6. Exopolymer analyses

Specific sugar residues in biofilm exopolysaccharide were de-tected using fluor-conjugated lectins (fluorescein isothiocyanate ortetramethyl rhodamine isothiocyanate (TRITC) - Sigma, St. Louis,Mo.) alone or in combination. Cy5 labeling was done using acommercial kit according to instructions provided (Research Or-ganics, Cleveland, Ohio). Specific lectins utilized included, Arachishypogaea (galactose, N-acetyl galactosamine), Canavalia ensiformis(mannose, glucose), Glycine max (N-acetyl galactosamine), Triticumestivum (N-acetyl glucosamine residues and oligomers), and Ulexeuropaeus (fucose). Image analyses and lectin binding volumecalculations were carried out using equations of Neu et al. (2001).

2.7. Carbon use spectra

Carbon-use spectra were determined using commercial Eco-plates (Biolog, Hayward, CA, USA) as described in Lawrence et al.(2004). Biofilm coupons (1�10 cm) were scraped with a sterilesilicon rubber spatula to remove biofilms and sonicated for 5 minin a Bransonic 5120 water bath sonifier (Branson Ultrasonics,Danbury, Conn.) to disperse cells. Each of the 96 wells of the Biologmicrotitre plates was subsequently inoculated with 150 mL of thebiofilm slurry and plates subsequently incubated at 23 ºC. Absor-bance (590 nm) was measured using a standard microtiter platereader at 24-h intervals for 7 d.

2.8. Protozoan and micrometazoan enumeration

Biofilm samples were removed from reactors weekly and pro-tozoan and micrometazoans counted manually on replicate 2 cm2

coupons (3 subsamples from each treatment replicate thereforen¼9 for each treatment) using phase contrast microscopy ac-cording to Packroff et al. (2002).

2.9. Molecular analyses

Bacterial cells in frozen biofilm samples were removed fromthe polycarbonate strip with a sterile metal scraper and total DNAextracted using the Powersoil DNA isolation kit (MO Bio Labora-tories, Inc., Carlsbad, CA.) according to the manufacturer’s in-structions with the following modifications. Samples were put intothe bead beater three times for 30 s at 5.5 m/s and cooled for10 min between each beating. Samples were cooled again beforecentrifugation.

2.10. Bacterial PCR amplification

A 584 bp fragment of the 16S rRNA Bacteria gene was amplifiedfor subsequent DGGE analysis. PCR amplification was conducted ina 50 mL reaction containing 1 mL of DNA template, 10 pmol of eachprimer, 1.25 U Taq DNA polymerase (Fermentas Canada, BurlingtonON) 1� PCR buffer, 2.5 mM MgCl2, and 200 mM dNTP (Table 1). Atouchdown PCR program using the PTC-200 thermocycler (Bio-radLaboratories, Mississauga, ON) consisted of an initial denaturationstep (94 °C for 5 min), followed by 10 cycles of denaturation (94 °Cfor 1 min), annealing at 66 °C (decreasing in each cycle by 1 °C) for

1 min and an elongation step (72 °C for 1 min). Then another 20cycles of 95 °C for 1 min, annealing at 56 °C for 1 min, and elon-gation at 72 °C for 1 min with a final elongation step of 72 °C for7 min. PCR product was confirmed by electrophoresis on a 1.5% w/v agarose gel in 1.0� Tris-acetate-EDTA (TAE) buffer (40 mM Tris,20 mM acetic acid and 1 mM EDTA) for 1.0 h at 100 V. Gels werestained using ethidium bromide and documented using the Al-phaImager 3300 gel documentation and image analysis system(Alpha Innotech Corporation, San Leandro, CA.).

2.11. Denaturing gradient gel electrophoresis analysis (DGGE)

After checking specificity and size of amplified products onagarose gels, the PCR product was separated by DGGE using anIngeny phorU2 system (Ingeny, Leiden, the Netherlands). Aliquotsof PCR product (45 μl) were mixed with 5 μL of loading dye bufferand resolved on a 6% (w/v) polyacrylamide gel in 1.0� TAE buffer,using denaturing gradients from 45% to 65% for Bacteria (100%denaturant contains 7 M urea and 40% deionised formamide).Electrophoresis was carried out at 60 V for 30 min, then 100 V for18 h at 60 °C. After electrophoresis, the gel was stained with SYBRgreen I (1,10 000 dilution; Molecular Probes, Eugene, OR) for15 min with gentle agitation and photographed using the AlphaI-mager 3300 gel documentation and image analysis system (AlphaInnotech Corporation, San Leandro, CA, USA).

2.12. Statistical analysis

Each antibiotic treatment had three identical reactors randomlyassigned to it on the reactor bench (replications). Furthermore,each analysis was done on subsamples of randomly selected bio-film coupons from among the 12 replicate coupons in each re-plicate reactor. In the biofilm experiments, a one way ANOVAfollowed by a post-hoc Tukey’s test was used to detect significantdifferences among sample means for all investigated variables atpo0.05 when the data were normally distributed (MiniTab, StateCollege, PA, USA).

For DGGE gels, band detecting, matching and processing werecompleted with the GelCompare II software 4.6 (Applied Maths,Kotrijk, Belgium). Fingerprint data were processed by generating aband matching table (Boon et al., 2002). The binary data wereexported and compared by principal component analysis (PCA)with PRIMER v6 software (PrimerE-Ltd Lutton, UK). Statisticalanalyses of PCA scores generated from the first two axes were runusing an analysis of similarity (ANOSIM) with PRIMERr v6 soft-ware (Clarke, 1993). The inclusion of DGGE ladders allowed

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–3934

GelCompareII to normalize the position of bands in all of the lanesunder examination.

3. Results

3.1. Biofilm thickness

Biofilm thickness in the TR and CL treatments in BEXP1 wasgreater than the control, while the ER treatment was significantly(po0.05) less (Fig. 1(A)). There was no significant difference be-tween the antibiotic nutrient controls and the control treatment.

In BEXP2 biofilm thickness in all ER treatments was less thanCL, TR, the antibiotic nutrient control and control (Fig. 1(B)), si-milar to results from BEXP1. Within ER treatments, thickness inthe 4 mg/L treatment was less than any of the other three con-centrations. Thickness in the CL treatment was less than thecontrol, while no difference was noted between the control andTR. TR was significantly greater than the comparative nutrienttreatment.

3.2. Biomass and production

In BEXP1, although no difference was noted in algal biomass(Chl a) when all antibiotic treatments were compared to thecontrol, bacterial biomass in the same treatments was significantly

Fig. 1. Influence of antibiotics on biofilm thickness in BEXP1 (A) and BEXP2 (B) asassessed at the microscale using confocal laser microscope analysis. Treatmentsindicated by different letters are significantly different from the control ( po0.05;n¼37SD).

Fig. 2. Results of image analyses of confocal laser micrographs illustrating the ef-fect of antibiotics on the relative abundance of algae, cyanobacteria, and bacteria inWascana Creek biofilms in BEXP1(A) and BEXP2 (B) Treatments indicated by dif-ferent letters are significantly different from the control (po0.05; n¼37SD).

less than the control (po0.05; Fig. 2(A)). A within antibiotictreatment comparison indicated that bacterial biomass was leastin the ER treatment, as was bacterial production (data not shown)while biomass in the CL treatment was greater than in the othertwo (po0.05). Bacterial production in the TR and CL treatmentwas not significantly different from the control (data not shown).The only treatment where there was an effect on cyanobacteriawas seen in the TR treatment (po0.05; Fig. 2(A)). In all treatmentsexcept ER, bacterial biomass was greater than that of algae in-dicating heterotrophy in these biofilms. The nutrient treatmentshad no significant effects relative to control and antibioticexposures.

In contrast to BEXP1, significant depression of algal biomass(from confocal laser analysis) was noted in TR and all ER treat-ments in BEXP2 when compared to the control (po0.05; Fig. 2(B)). Within ER treatments, algal biomass in the 3 and 4 mg/Ltreatments was less than in 1 and 2 mg/L while biomass in 2 mg/Lwas less than in 1 mg/L (po0.05). Bacterial biomass was also af-fected being far less in the TR and all ER treatments when com-pared to either CL or the control. Within ER treatments, bacterialbiomass in the 1 and 4 mg/L treatments were greater than the re-maining two, while biomass in the 4 mg/L was less than that in the1 mg/L (po0.05). The 4 mg/L nutrient control resulted in increasedbacterial biomass relative to CL, while cyanobacteria and bacterialbiomass increased relative to TR. With regard to ER 1 mg/L bacterialbiomass increased relative to the nutrient addition, while for the 2,3, and 4 mg/L treatments algal biomass was significantly reducedrelative to the nutrient addition (Fig. 2(B)). In contrast to BEXP1, inall treatments, algal biomass was greater than that of bacteria,indicating autotrophy. All antibiotic treated biofilms, except TR hadcyanobacterial biomass which was less than the control (Fig. 2(B)).No effect of any antibiotic on bacterial production was noted (datanot shown).

Fig. 3. Results of image analyses of confocal laser micrographs illustrating the ef-fect of antibiotics on the composition of Wascana Creek biofilm exopolymer fromBEXP1 (A) and BEXP2 (B). A panel of 5 lectins was used to assess the nature of theglycoconjugates present in the biofilms. Treatments indicated by different lettersare significantly different from the control (po0.05; n¼37SD).

Fig. 4. (A,B,C). Differential display of the carbon utilization assays of control, nu-trient control and clindamycin (CL - A), erythromycin (ER - B) and trimethoprim (TR- C) treated reactors from BEXP1. Data points in circles are significantly differentfrom their respective control value at a level of po0.05.

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–39 35

3.3. Lectin binding volume

In BEXP1, the least lectin binding volume was observed in theER treatment, while the most was in the TR treated biofilms (Fig. 3(A)). In all cases the respective nutrient controls differed sig-nificantly from the control. CL and TR treatments were also sig-nificantly different from their nutrient controls. In the case of ER,however, lectin binding volumes were not significantly differentbetween antibiotic and nutrient treatments (Fig. 3(A)). In BEXP2slightly different results were noted in that biofilms from all drugtreatments had less lectin binding volume than the control (Fig. 3(B)). Similar to BEXP1, however, the least lectin binding volumewas observed in the ER treatments. Additionally, no significantdifference was observed between the four ER treatments.

3.4. Carbon utilization spectra

In BEXP1, significant differences were noted when antibioticnutrient control and antibiotic treatments were compared to thecontrols in all carbon categories except phosphorylated com-pounds. In most cases, antibiotics significantly decreased utiliza-tion of the tested carbon substrates (Fig. 4(A),(B),C). Of the numberof significant effects noted, decreased utilization ranged from 64%in CL treatments to 100% for both ER and TR. Interestingly, theaddition of the antibiotic nutrient equivalents similarly resulted ina general trend of decreased carbon substrate utilization. Due totechnical difficulties (spectrophotometer not zeroing correctly) thecarbon utilization data from BEXP2 could not be utilized.

3.5. Protozoan and micrometazoan enumeration

By the end of BEXP1 and BEXP2, protozoan and rotifer numberswere higher in the TR and lower in ER treatment when compared

to the control (Fig. 5(A)). In contrast, although protozoan numbersin the CL treatment in BEXP1 were less than the control, rotifernumbers were greater on the final day of the experiment. Resultsfor the CL treatment in BEXP2 were somewhat different withprotozoan and rotifer numbers less than the control by experimentend (Fig. 5(B)).

Within the ER dose response treatments, numbers of proto-zoans and rotifers were slightly higher in the 1 mg/L treatment,protozoans were similar to the control in the 3 mg/L treatment,while protozoans and rotifers in the remaining treatments wereless than the control (Fig. 5(B)). Oddly, the ER nutrient addition of4 μg/L resulted in the least numbers of both protozoans and

Fig. 5. Cumulative enumeration of protozoa (ciliates, flagellates, amoeba) and micrometazoa (rotifers, nematodes) from BEXP1 (A) and BEXP2 (B) showing the effects ofantibiotic exposure on grazer density within Wascana biofilm communities.

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–3936

microinvertebrates by experiment end (Fig. 5(B)).

3.6. Molecular analyses- PCA and Anosim results

3.6.1. BEXP1Principal component analyses of the DGGE results indicated

that each antibiotic treatment resulted in significantly differentcommunities. The control and CL treatments grouped together,while the ER and TR treatments were different from each other aswell as from the CL/control grouping (Fig. 6(A)). Further the TRtreatment grouped not only with its nutrient control but the othernutrient controls as well. ANOSIM pair-wise analysis showed sig-nificant differences (R¼0.799, po0.001%) in the structure of

microbial communities between antibiotic treatments. The MSDSbelow 0.2 indicated that the graph was highly interpretable andthe model statistically significant. According to this analysis, CLwas the only treatment not significantly different from the control(Fig. 6(A)). Bacterial diversity in the remaining treatments (TR andER) was significantly different from the control. The greatest effectwas observed in the ER treatment with diversity here groupingaway from both control and all other treatments (Fig. 6(A)). Theapparent similarity between ER and CL nutrient control was likelydue to the outlier and does not necessarily reflect a true similarity.

3.6.2. BEXP2DGGE principle components analysis indicated as in BEXP1 that

Fig. 6. Results of PCA on replicate DGGE analyses showing the significant differ-ences (po0.001 ANOSIM) between ER and TR-treated communities relative to thecontrol in BEXP1 (A) and BEXP2 (B).

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–39 37

CL and control communities grouped together. As well the 1, 2 and3 mg/L ER treatments assembled together. The TR and 4 mg/L ERgrouped together but separate from the other treatments (Fig. 6(B)). The Global R value for the ANOSIM was 0.93 at po0.001%. Asin BEXP1, bacterial diversity in the CL treatment was not differentfrom the control. ER4 and TR grouped together indicating moresimilarity to each other than to the other treatments (Fig. 6(B)).The clustering of the remaining ER treatments suggested theirbanding patterns were more similar to one another than whencompared to the ER4/TR, ER nutrient control or CL/control.

4. Discussion

It has been stated that the ability of biofilms to detect toxiceffects of anthropogenic stressors both in the short (changes inphysiological function) and long term (changes in communitystructure) make them ideal early indicators of stress in aquaticecosystems (Sabater et al., 2007; Sandin et al., 2009; Lawrenceet al., 2015). It is evident from the results of the Wascana experi-ments presented here just how ideal biofilms can be as such in-dicators. These results clearly show that the test antibiotics (ER, TRand CL) changed the amount of bacterial and algal biomass pro-duced, carbon utilization patterns, microbial community compo-sition and structure, as well as the amount and composition ofexopolysaccharide created in stream biofilms.

One of the first observations was the significant depression ofalgal biomass in all ER treatments in the second experiment(BEXP2). This observation contrasted with results of the first ex-periment (BEXP1) where no depression was noted. There are anumber of explanations for this disparity. It may be that differentalgal communities, with differing sensitivities to antibiotics, arerecruited to biofilms during warmer August as opposed to cooler

November temperatures. In Wascana Creek, water temperaturesvary from about 21 °C in July to 5 °C in November (Waiser et al.,2011a). Research into diclofenac effects on South SaskatchewanRiver biofilms, has noted similar effects with spring and summerbiofilms differing in their microbial communities and response(Lawrence et al., 2007). According to the authors, seasonal factorslike flow rate, water temperature and levels as well as amount ofphotosynthetically active radiation likely influenced microorgan-isms available for recruitment to the biofilm (Lawrence et al.,2007; Besemer et al., 2007).

There are, however, other explanations for the declines inbiofilm algal and bacterial communities. Although depression ofalgal biomass in the ER treatment only occurred in BEXP2, bac-terial biomass was always significantly less than the control nomatter what the season, or ER concentration. Erythromycin andother macrolide antibiotics inhibit bacterial protein synthesis bybinding to the 23S rRNA molecule (in the 50S subunit) of thebacterial ribosome blocking exit of the growing peptide. Declinesin bacterial biomass may therefore have been the result of thedirect action of ER on bacteria protein production. Depression ofalgal biomass in BEXP2 may also have been due to the coherenceof green algae phylogeny with plastid genes containing consensusprokaryotic promoters and bacterial-like sigma factors (Brain et al.,2008). Because of the existence of these transcription/translationhomologues, macrolide antibiotics, like ER, may disturb the tran-scription/translation process in the photosynthetic chloroplast(Brain et al., 2008) thereby lowering algal biomass.

But a depression in algal biomass was also noted in the TRtreatment of BEXP2. The basis of TR toxicity appears to be itsability to bind and reversibly inhibit dihydrofolate reductase, anenzyme involved in folate synthesis (Crane et al., 2006). Essentialfor one-carbon transfer reactions in plants, folates are also neededfor photorespiration and production of nucleic acids, amino acidsand lignins (Brain et al., 2008). Because the folate syntheticpathway in plants and bacteria is essentially the same, TR couldpotentially elicit a physiological response in phytoplanktonthereby providing a possible explanation for the depressed algalbiomass noted here.

Additionally, the decline in bacterial biomass may also be re-lated to the depression in algal biomass. Within biofilms, algaehave been referred to as ‘ecosystem engineers’, physically mod-ifying the microenvironment and forming a structural template onwhich bacteria can grow (Besemer et al., 2007). They also excretelabile organic carbon that can be readily used by bacteria forgrowth. As such they play an important role in biofilm structureand function. In BEXP2, the depression in algal biomass may havebeen accompanied by a decline in excreted labile carbon. Thisdecline in turn may have been linked to the concomitant changesnoted in the bacterial communities. Within aquatic ecosystems thebacterial constituents are also important in conditioning detritus,making it more palatable for grazers (Maul et al., 2006). Conse-quently if there are less ‘conditioning’ bacteria, there might be aspill down effect on those grazers relying on biofilm bacteria toperform this function.

Within aquatic ecosystems, bacteria and algae both play keyroles in the biogeochemical cycling of carbon, nitrogen andphosphorus. Disruption of their communities by exposure to an-tibiotics may affect these important cycles (Rosi-Marshall andRoyer, 2012). In the Wascana studies presented here biofilm car-bon utilization patterns were affected with significantly decreaseduse of some carbon substrates observed in all antibiotic treat-ments. This observation suggests that some disruption of bacterialcarbon cycling may be occurring as a result of antibiotic exposure,a similar result to those noted in other studies. Using Biolog Ecoplates, stream microbial communities exhibited declines in carbonutilization patterns after exposure to 100 mg/L ciprofloxacin (Maul

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–3938

et al., 2006), while Lawrence et al. (2009) found that exposure to10 mg/L of triclocarban and triclosan also resulted in depressedcarbon utilization.

There are a number of possible explanations for the observa-tion of depressed carbon utilization. Results of the PCA analysis onthe DGGE gels clearly showed that the microbial communities inTR and ER exposed communities were different from the control.The extent to which these changes affected carbon cycling is un-known, but remains a possibility. Additionally, some studies havedemonstrated that antibiotics act as metabolic signaling moleculeswith bacteria responding to exposure by up or down regulation ofgenes (Goh et al., 2002; Fajardo and Martínez, 2008). Disruption ofcarbon cycling or other cellular functions may occur as a result. Atexposures of 50 and 500 mg/mL of ER, for example, activation orrepression of a significant proportion of genes (�5%) involved intransport, virulence and DNA repair were noted in Escherichia coliand Salmonella typhimurium (Goh et al., 2002). With respect toWascana Creek, a recent metagenomics and meta-transcriptomicstudy (using the same reactor approach and water source as in thisstudy) investigated the effects of ER (at 1 mg/L) on Wascana Creekbiofilms. Here, active community composition at the RNA level(taxonomic affiliation of mRNA sequencing) in Cyanobacteria andProteobacteria was strongly affected by ER exposure (Yergeauet al., 2012). Effects on photosynthesis-related genes were alsonoted (Yergeau et al., 2012). When compared to the control, ERexposure caused expression of several unique genes which re-sulted in large changes in metabolic networks specifically thoserelated to C, N and P cycling. Such observed genetic regulatorychanges may explain the decreased carbon utilization noted in thepresent study.

A number of studies have indicated that decreased carbonsubstrate utilization influences biofilm exopolysaccharide (EPS)composition (Mohamed et al., 1998; Lawrence et al., 2007, 2015).In the Wascana experiments presented here, definite changes inEPS composition were observed in antibiotic exposed biofilms.Such changes are of concern because the EPS matrix is responsiblefor maintaining cell spatial organization and promoting cross-linking or scaffolding within the biofilm, thereby fostering growthof bacterial microcolonies. As well, some EPS’s make biofilms moreviscous thereby favouring biofilm spreading (Chew et al., 2014).Decreased carbon utilization may also affect the amount of EPSproduced (Lawrence et al., 2007, 2015).

Results from the Wascana experiments clearly indicated thatantibiotic exposure, especially to ER also resulted in thinner bio-films and this observation is of concern. According to some au-thors, physiological changes, including declines in biofilm thick-ness, are indicative of an initial stress response and may be anearly sign of direct ecosystem damage (Bonnineau et al., 2010). Instream and river ecosystems biofilms integrate nutrients and or-ganic matter into a structure subsequently utilized as a foodsource to support higher trophic levels (invertebrate grazers).Biofilm ingestion is also thought to provide endosymbionts whichexcrete enzymes (apparently lacking in some invertebrate grazers)capable of breaking down cellulose and other refractory leaf con-stituents (Maul et al., 2006; Sinsabaugh et al., 1985). Less biofilm ofdiffering composition (as noted in the present study) and perhapsof less nutritional value may have negative consequences for in-vertebrate grazers. If the negative effect is great enough or occursoften enough it may resonate to higher trophic levels. In theWascana experiments there was an indication that such resonancemight be occurring as protozoan and metazoan numbers in mostof the ER treatments and rotifers in the higher ER treatments weresignificantly lower than the control by experiment end.

Although decreased carbon utilization may affect the amountof EPS produced, there may be other factors which influence thisprocess. Recent research has suggested that antibiotic exposure

may affect cell surfaces, making bacterial adhesion more difficult.ER at concentrations of 0.5 and 50 mg/L, for example, significantlyinhibited adhesion of Escherichia coli, Bacillus subtilis and Aqua-bacterium commune, to uncoated polystyrene beads (Schreiberet al., 2008).

One of the interesting and perplexing observations from BEXP1was the significant difference not only in biomass amount but EPScomposition in both the CL and ER nutrient controls when com-pared to the creek water only control. One reasonable explanationwould be that nutrient addition in these two treatments may havecaused an initial rapid growth in both biofilm bacteria and algae.This rapid growth may have been followed by an increase inprotozoan numbers and subsequent grazing pressure. The grazingeffort in turn may have been enough to contribute to the differ-ences in biofilm biomass and EPS composition noted in BEXP1.There is some support for this idea in that at experiment end,highest protozoan numbers were noted in the ER and Cl nutrientcontrols. Research has shown that the structure and production ofperiphyton (biofilms) communities can be regulated by such in-terplay between consumers and nutrients (Liess and Hillebrand,2006). Unfortunately we do not have a time series of biomass orEPS composition to indicate the initial response of biofilms to thenutrient treatment. Other biofilm experiments, however, haveshown that phosphorus additions were responsible for increasedbiofilm bacterial growth and concomitant increase in protistsnumbers. Protist grazing in turn rapidly reduced the amount ofextracellular polymeric substances accumulated by biofilm bac-teria (Mohamed et al., 1998).

The same explanation, however, does not hold for BEXP2. Herealthough the biomass in the nutrient treatment was significantlyless than the control, protozoan and metazoan grazer numberswere the lowest of all treatments. Research has shown that com-munity composition within biofilms changes seasonally. In 2005,for example, a study of the seasonal response of biofilms to dis-solved organic matter and nutrient enrichments in the MahoningRiver, NE Ohio revealed that the relative abundance of differingBacteria phylogenetic groups varied seasonally and with nutrienttreatment. Specifically, bacterial community structure of biofilmsgrown in the fall and exposed to phosphate enrichment was sig-nificantly different from the control (Olapade and Leff, 2005). Thesecond biofilm experiment was conducted with water collectedfrom Wascana Creek in November. Results from the BEXP2 DGGEPCA analyses clearly indicate that the bacterial community grownin the nutrient amendment was significantly different from thecontrol as well as the antibiotic treatments. The dissimilarity in thebacterial community noted in addition to the time of year in whichwater was collected, may have contributed to the differences no-ted between the nutrient control and the control.

5. Conclusion

Taken together, results of this study represent an importantstep in our understanding of the effects of antibiotics (especiallyER) on attached stream microbial communities. They providestrong evidence that chronic exposure to antibiotics at en-vironmentally relevant concentrations can cause changes not onlyin the amount of EPS produced and its composition but theamount of algal and bacterial biomass produced. Changes in mi-crobial community composition as well as some disruption incarbon utilization as also occurred as a result of antibiotic ex-posure. Knowledge gained from this research is also importantbecause ER is ubiquitous– found worldwide in almost every stu-died waterbody receiving treated sewage effluent. The WascanaCreek research presented here suggests that ER may be re-sponsible, at least in part, for observed ecosystem disturbance in

M.J. Waiser et al. / Ecotoxicology and Environmental Safety 132 (2016) 31–39 39

other waterbodies receiving treated sewage effluent.

Acknowledgements

Funding for this research was provided by Health Canada andEnvironment Canada.

Appendix A. Supplementary material

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ecoenv.2016.05.026.

References

Ashton, D., Hilton, M., Thomas, K.V., 2004. Investigating the environmental trans-port of human pharmaceuticals to streams in the United Kingdom. Sci. TotalEnv. 222, 167–184.

Backhaus, T., Porsbring, T., Arrhenius, A., Brosche, S., Johansson, P., Blanck, H., 2011.Single substance and mixture toxicity of five pharmaceuticals and personal careproducts to marine phytoplankton communities. Env. Toxicol. Chem. 30, 1–11.

Besemer, K., Singer, G., Limberger, R., Chlup, A.K., Hochedlinger, G., Hödl, I., Baranyi,C., Battin, T.J., 2007. Biophysical controls on community succession in streambiofilms. Appl. Env. Microbiol. 73, 4966–4974.

Brain, R.A., Hanson, M.L., Solomon, K.R., Brooks, B.W., 2008. Aquatic plants exposedto pharmaceuticals. Effects and risks. Rev. Env. Contam. Toxicol. . 192, 67–115.

Boon, N., Windt, W.D., Verstraete, W., Top, E.M., 2002. Evaluation of nested PCRDGGE (denaturing gradient gel electrophoresis) with group-specific 16S rRNAprimers for the analysis of bacterial communities from different wastewatertreatment plants. FEMS Microbiol. Ecol. 39, 101–112.

Bonnineau, C., Guasch, H., Proia, L., Ricart, M., Geiszinger, A., Romaní, A.M., Sabater,S., 2010. Fluvial biofilms, a pertinent tool to assess b-blockers toxicity. Aquat.Toxicol. 96, 225–233.

Brosius, J., Dull, T.J., Sleeter, T.J., Noller, D.D., H, F., 1981. Gene organization andprimary structure of a ribosomal RNA operon from Escherichia coli. J. Bacteriol.148, 107–127.

Casamayor, E., Schafer, H., Baneras, L., Pedros-Alio, C., Muyzer, G., 2000. Identifi-cation of and spatio-temporal differences between microbial assemblages fromtwo neighboring sulfurous lakes: comparison by microscopy and denaturinggradient gel electrophoresis. Appl. Env. Microbiol. 66, 499–508.

Castiglioni, S., Pomati, Y., Miller, K., Burns, B.P., Zuccato, E., Calamari, D., Neilan, B.A.,2008. Novel homologs of the multiple resistance regulator marA in antibioticcontaminated environments. Wat. Res. 42, 4271–4280.

Chew, S.C., Kundukad, B., Seviour, T., van der Maarel, J.R.C., Yang, L., Rice, S.A., Doyle,P., Kjelleberg, S., 2014. Dynamic remodeling of microbial biofilms by function-ally distinct exopolysaccharides. 5 (4). http://dx.doi.org/10.1128/mBio.01536-14.

Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in communitystructure. Aust J. Ecol. 18, 117–143.

Crane, M., Watts, C., Boucard, T., 2006. Chronic aquatic environmental risks fromexposure to human pharmaceuticals. Sci. Total Env. 367, 23–41.

Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products inthe environment, agents of subtle change? Env. Health Persp. 107, 907–938.

Eggen, R.I.L., Behra, R., Burkhardt-Holm, P., Escher, B.I., Schweigert, N., 2004.Challenges in ecotoxicology. Env. Sci. Technol. 38, 58A–64A.

Fajardo, A., Martínez, J.L., 2008. Antibiotics as signals that trigger specific bacterialresponses. Curr. Opin. Microbiol. 11, 161–167.

Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of human pharmaceu-ticals. Aquat. Toxicol. 76, 122–159.

Goh, E., Yin, G., Tsui, W., McClure, J., Surette, M.G., Davies, J., 2002. Transcriptionalmodulation of bacterial gene expression by subinhibitory concentrations ofantibiotics. Proc. Natl. Acad. Sci. 99, 17025–17030.

Gullberg, E., Cao, S., Berg, O.G., Ilback, C., Sandegren, L., 2011. Selection of resistantbacteria at very low antibiotic concentrations. PLoS Pathog. 7 (7). http://dx.doi.org/10.1371/journal.ppat.1002158.

Johnson, A.C., Keller, V., Dumont, E., Sumpter, J.P., 2015. Assessing the concentra-tions and risks of toxicity from the antibiotics ciprofloxacin, trimethoprim anderythromycin in European rivers. Sci. Total Env. 511, 747–755.

Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2004. Potential ecological and humanhealth risks associated with the presence of pharmaceutically active com-pounds in the aquatic environment. Crit. Rev. Toxicol. 34, 335–350.

Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.,Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewatercontaminants in U.S. streams 1999–2000, a national reconnaissance. Env. Sci.Technol. 36, 1202–1211.

Lawrence, J.R., Chénier, M., Roy, R., Beaumier, D., Fortin, N., Swerhone, G.D.W., Neu,T.R., Greer, C.W., 2004. Microscale and molecular assessment of the impacts of

nickel, nutrients, and oxygen level on river biofilm communities. Appl. Env.Microbiol. 70, 4326–4339.

Lawrence, J.R., Swerhone, G.D.W., Wassenaar, L.I., Neu, T.R., 2005. Effects of selectedpharmaceutical s on riverine biofilm communities. Can. J. Microbiol. 51,655–669.

Lawrence, J.R., Swerhone, G.D.W., Topp, E., Korber, D.R., Neu, T.R., Wassenaar, L.,2007. Structural and functional responses of river biofilm communities to thenonsteroidal anti-inflammatory diclofenac. Env. Toxicol. Chem. 26, 573–582.

Lawrence, J.R., Zhu, B., Swerhone, G.D.W., Roy, J., Wassenaar, L.I., Topp, E., Korber, D.R., 2009. Comparative microscale analysis of the effects of ticlosan and triclo-carban on the structure and function of river biofilm communities. Sci. TotalEnv. 407, 3307–33016.

Lawrence, J.R., Zhu, B., Swerhone, G.D.W., Roy, J., Tumber, V., Waiser, M.J., Topp, E.,Korber, D.R., 2012. Molecular and microscopic assessment of the effects ofcaffeine, acetaminophen, diclofenac and their mixtures on river biofilm com-munities. Env. Toxicol. Chem. 31, 508–517.

Lawrence, J.R., Topp, E., Waiser, M.J., Tumber, V., Roy, J., Swerhone, G.D.W., Leavitt,P., Paule, A., Korber, D.R., 2015. Resilience and recovery, the effect of triclosanexposure timing during development, on the structure and function of riverbiofilm communities. Aquat. Toxicol. 161, 253–266.

Liess, A., Hillebrand, H., 2006. Role of nutrient supply in grazer-periphyton inter-actions: reciprocal influence of periphyton and grazer nutrition stoichiometry.J. N. Am. Benthol. Soc. 25, 632–642.

Maul, J.D., Schuler, L.J., Belden, J.B., Whiles, M.R., Lydy, M.J., 2006. Effect of theantibiotic ciprofloxacin on stream microbial communities and detrivorousmacroinvertebrates. Env. Toxicol. Chem. 25, 1598–1606.

Mohamed, M.N., Lawrence, J.R., Robarts, R.D., 1998. Phosphorus limitation of het-erotrophic biofilms from the Fraser River, British Columbia, and the effect ofpulp mill effluent. Microb. Ecol. . 36, 121–130.

Morgan, S., McMahon, M., Lam, J., Mooney, D., Raymond, C., 2005. The Canadian RxAtlas. University of British Columbia. Centre for Health Services and PolicyResearch.

Muyzer, G., De Waal, E.C., Uitterlinden, A.G., 1993. Profiling of complex microbialpopulations by denaturing gradient gel electrophoresis analysis of polymerasechain reaction-amplified genes coding for 16S rRNA. Appl. Env. Microbiol. 59,695–700.

Neu, T.R., Wölfl, S., Lawrence, J.R., 2004. 3-dimensional differentiation of photo-trophic biofilm constituents by multi channel confocal and 2-photon laserscanning microscopy. J. Microbiol. Meth. 56, 161–172.

Neu, T.R., Swerhone, G.D.W., Lawrence, J.R., 2001. Assessment of lectin-bindinganalysis for in situ detection of glycoconjugates in biofilm systems. Micro-biology 147, 299–313.

Olapade, O.A., Leff, L.G., 2005. Seasonal responses of stream biofilm communities todissolved organic matter and nutrient enrichments. Appl. Env. Microbiol. 71,2278–2287.

Packroff, G., Lawrence, J.R., Neu, T.R., 2002. In situ confocal laser scanning micro-scopy of protozoans in pure cultures and complex communities. Acta Protozool.41, 245–253.

Pomati, F., Netting, A.G., Calamari, D., Neilan, B.A., 2004. Effects of erythromycin,tetracycline and ibuprofen on growth of Synechocystis sp. and Lemna minor.Aquat. Toxicol. 67, 387–396.

Robarts, R.D., Wicks, R.J., 1989. [Methyl-3H]thymidine macromolecular incorpora-tion and lipid labeling, Their significance to DNA labeling during measurementsof aquatic bacterial growth rate. Limnol. Oceanogr. 34, 213–222.

Rosi-Marshall, E.J., Royer, T.V., 2012. Pharmaceutical compounds and ecosystemfunction: an emerging research challenge for aquatic ecologists. Ecosystems 15(6), 867–880.

Sabater, S., Guasch, H., Ricart, M., Romani, A., Vidal, G., Klüner, C., Schmitt-Jansen,M., 2007. Monitoring the effect of chemicals on biological communities, thebiofilm as an interface. Anal. Bioanal. Chem. 387, 1425–1434.

Sandin, L., Solimini, A.G., 2009. Freshwater ecosystem structure-function relation-ships, from theory to application. Fresh Biol 54, 2017–2024.

Schreiber, F., Szewzyk, U., 2008. Environmentally relevant concentrations of phar-maceuticals affect bacterial adhesion. Aquat. Toxicol. 87, 227–233.

Sinsabaugh, R.L., Linkins, A.E., Benfield, E.F., 1985. Cellulose digestion and assim-ilation by three leaf-shredding aquatic insects. Ecology 66, 1464–1471.

Waiser, M., Robarts, R.D., 1997. Impacts of a herbicide and fertilizers on the mi-crobial community of a saline prairie lake. Can. J. Fish. Aquat. Sci. 54, 320–329.

Waiser, M.J., Tumber, V., Holm, J., 2011a. Effluent dominated streams Part I, pre-sence and effects of excess nitrogen and phosphorus in Wascana Creek, SK. Env.Toxicol. Chem. 30, 496–507.

Waiser, M.J., Humphries, D., Holm, J., Tumber, V., 2011b. Effluent dominatedstreams Part II, Presence and possible risk of pharmaceuticals and personal careproducts in Wascana Creek, SK. Env. Toxicol. Chem. 30, 508–519.

Zuccato, E., Castiglioni, S., Fanelli, R., Reitano, G., Bagnati, R., Chiabrando, C., Pomati,F., Rosetti, C., Calamari, D., 2006. Pharmaceuticals in the environment in Italy;causes, occurrence, effects and control. Env. Sci. Pollut. Res. 13, 15–21.

Yang, L.-H., Ying, G.-G., Su, H.-C., Stauber, J.L., Adams, M.S., Binet, M.T., 2008.Growth-inhibiting effects of 12 antibacterial agents and their mixtures on thefreshwater microalga Pseudokirchneriella subcapitata. Env. Chem. 27,1201–1208.

Yergeau, E., Sanschagrin, S., Waiser, M.J., Lawrence, J.R., Greer, C.W., 2012. Sub-in-hibitory concentrations of different pharmaceutical products affect the meta-transcriptome of river biofilms. Env. Microbiol. Rep. 4, 350–359.