DESALINATION

www.elsevier.com/locate/desalDesalination 247 (2009) 170–179

Removal of bacterial and viral indicators in vertical flowconstructed wetlands and intermittent sand filters

Antonina Torrensa*, Pascal Molleb, Catherine Boutinb, Miquel Salgota

aSoil Science Laboratory, University of Barcelona, Avda. Joan XXIII s/n, 08028 Barcelona, SpainbCemagref, Water quality and pollution prevention research team, 3 bis quai chauveau

CP 220 69336 Lyon Cedex 09, FranceTel: + 34 (93) 402-4494; Fax: + 34 (93) 402-4495; email: antoninatorrens@ub.edu

Received 20 September 2007; revised 12 February 2008; accepted 10 March 2008

Abstract

The removal of faecal coliforms (FC), Escherichia coli, somatic coliphages and F-specific bacteriophages wasmonitored over a period of 2 years in two vertical flow constructed wetlands (VFCWs) and four intermittent sandfilters (ISFs) that treat pond effluent. Tracer tests (NaCl) were performed to estimate the hydraulic retention time(HRT) of the filters. The contribution of Phragmites australis to the removal of the faecal indicators was exam-ined. The effect of media characteristics (depth, type of sand), hydraulic load and water application schedule ondisinfection was also addressed. Planted and non-planted vertical sand filters were equally effective in removingmicroorganisms. The disinfection capacity was very sensitive to the depth and operation of the filters. HRT wasthe key parameter regarding filter disinfection capacity. Bacterial indicators were removed at a higher rate thanviral ones. Somatic coliphages were removed at higher rates than F-specific bacteriophages.

Keywords: E. coli; Faecal coliforms; Hydraulic retention time; Phages; Sand filters; Wetlands

1. Introduction

The capacity of subsurface horizontal flowconstructed wetlands (HFCWs) to remove indica-tor microorganisms, mainly bacteria indicators,has been thoroughly examined [1–3]. However,few studies have addressed the disinfection effec-tiveness of vertical flow constructed wetlands(VFCWs). Furthermore, the conclusions drawn

by studies on HFCWs regarding the mechanismsinvolved in the removal of microorganisms ordesign recommendations cannot be extrapolatedto vertical systems because the hydraulics pat-terns and oxygen conditions of these two kindsof wetlands differ greatly. The performance ofVFCWs would resemble that of infiltration-percolation systems.

The removal of microorganisms during infil-tration in porous media is normally attributed tothe combination of straining, adsorption and*Corresponding author.

Presented at Multi Functions of Wetland Systems, International Conference of Multiple Roles of Wetlands, June 26–29, 2007, Legnaro (Padova) Italy

0011-9164/08/$– See front matter � 2009 Elsevier B.V. All rights reserved

inactivation. The efficiency of these processes isrelated to several factors. Straining is influencedby the physical characteristics of the filtermedium, hydraulic loading and clogging [4].Adsorption is controlled mainly by the grain sur-face characteristics of the porous medium, waterflow velocity, wastewater ionic strength, pH,moisture content and cell surface characteristics[5]. Microbial die-off depends on temperatureand the presence of other microorganisms in theporous media, especially grazing protozoa [6].The efficiency of most of these processes is there-fore related to physical flow conditions, and stud-ies on the removal of microorganisms shouldconsider all the parameters that affect hydraulicsduring infiltration in porous media (e.g., filteringmaterial, filter depth, hydraulic load (HL),hydraulic retention time (HRT) and feeding sched-ule). In this regard, several authors have describedthe relationship between the removal of bacterialindicators and water retention time in intermittentsand filters (ISFs) [7,8] and HFCWs [9] using con-servative tracer experiments.

The contribution of macrophytes to theremoval of faecal indicators remains unclear.A number of studies report that bacterial popula-tions are enhanced in the rhizosphere, as macro-phytes offer a large surface area for thedevelopment of attached biomass [10]. Accord-ing to Decamp et al. [11], the improved aerationof the root zone of the planted filters in compar-ison to non-planted ones provides an additionalexplanation for the higher removal efficiencyof the former. Moreover, some authors reportthat root exudates produce bactericidal substan-ces [12]. All these effects contribute to the elim-ination of pathogen microorganisms. In thisregard, a higher removal efficiency has beendescribed for planted filters [6,13,14]. Wand etal. [6] concluded that predation is the dominantmechanism of bacteria removal in planted andnon-planted soil filters and constructed wet-lands. They attributed the higher removal effi-ciencies of the planted filters to a higher

protozoa content. The improved aeration of theroot zone in planted gravel beds would providebetter conditions for protozoa development. Incontrast, other studies report no significantdifference in the performance of planted andnon-planted systems [15,16]. In addition,Vacca et al. [17] report that bacteria removalin vertical sand filters planted with Phragmitesaustralis is lower than that in non-planted ones.

Few studies have examined the fate of virusesand viral indicators during subsurface flow wet-land treatment. The inactivation rates of theviral indicators in constructed wetlands and infil-tration percolation systems differ from that ofconventional bacterial indicators [18,19].

The purpose of this work was to examine thefate of bacterial and viral indicators in plantedand non-planted vertical sand filters. Theremoval of faecal coliforms (FC), Escherichiacoli, somatic coliphages and F-specific bacterio-phages was monitored over a period of 2 years in2 VFCWs and 4 ISFs treating a facultative wasstabilization pond effluent. The contribution ofPhragmites australis to the elimination ofthese pathogen indicators was examined. Inaddition, the effect of the filtering bed character-istics (filter depth, type of sand), hydraulic load,frequency of feeding and temperature on the dis-infection capacity of the filters was also investi-gated. Tracer tests (NaCl) were carried out toestablish the relation between microorganismremoval and the hydraulic pattern of the filters.

2. Material and methods

2.1. Experimental design

Effluent from a facultative pond located inAurignac (France) was applied to six filters ofdistinct designs, each with a surface of 50m2

(1m2/PE). Three types of filtering mediumwere tested (river sand, river sand with Phrag-mites australis and crushed sand). Two filterdepths were used for each type of medium:

A. Torrens et al. / Desalination 247 (2009) 170–179 171

25 and 65 cm. The particle-size distribution ofthe sand was chosen following the recommenda-tions described by Lienard et al. [20]. Themain characteristics of the filters are shown inTable 1.

Filters were intermittently fed: floodingsequences alternating with drainage sequences.The daily number of feeding-drainage cycles (f )ranged from 4 to 32 depending on the HL appliedand the unit dose, i.e. the water height appliedduring each feeding sequence (Table 2). The fil-ters were alternately fed for 3–4 days followedby a rest period of 7 days.

2.2. Tracer tests

Twenty tests were carried out in all filtersusing sodium chloride (NaCl), the concentration

of which can be easily monitored by electricalconductivity sensors. In each test, the tracer wasadded to the water of only one feeding sequence.The tracer content and the flow rate at the outletof the filters were recorded at a one minute timestep. This monitoring allowed to determine thedetention time distribution (DTD), DTD = E(t),and the mean HRT. In addition, a minimumHRT was calculated as the time required toreach 10% of initial salt concentration in thefilter effluent. This parameter was reported byStevik et al. [21] to be more relevant for predict-ing microbiological removal than mean HRT.

2.3. Microbiological monitoring

The analysis of 24-h composite samples (30campaigns) and grab samples (collected every

Table 1

Main characteristics of the filter

Sand Characteristics

Filter Depth (cm) Filtering medium d102 CU3

Fines content(% by weight)

Porosity(%)

M25 25 Planted1 river sand0.25 4.7 2.1 0.42

M65 65 Planted1 river sand

R25 25 River sand0.25 4.7 2.1 0.42

R65 65 River sand

C25 25 Crushed sand0.19 9.3 4.0 0.44

C65 65 Crushed sand

1Reeds: Phragmites australis,2d10: Mesh diameter allowing 10% of the sand mass to pass through (mm).3CU: Coefficient of uniformity: ratio d60/d10.

Table 2

Operation stages

Feeding modes

Stage Duration (months) Hydraulic load (cm/day) Unit dose (cm) Number of batches per day

1 4 20–25 5 4–52 10 35–40 5 7–83 5 75–80 5 15–164 5 75–80 2.5 30–32

172 A. Torrens et al. / Desalination 247 (2009) 170–179

2 to 3 weeks) was performed from September2003 to October 2005. Samples were takenfrom the influent and the effluent of the filtersand then preserved and stored following thestandard methods [22]. The analysis of themicrobiological indicators (FC, E. coli, somaticcoliphages and F-specific bacteriophages) wasperformed within 24 h, following the methodsdescribed in Table 3.

2.4. Temperature monitoring

Water temperature was measured in each sam-ple with a portable thermometer. Temperatureinside the filters was also monitored by six sensors(1 per filter) located at a depth of 15 cm. Temper-atures were continuously recorded (every minute)from September 2003 to September 2005.

2.5. Data analyses

Correlation and covariance analyses wereused to test the relationship between the varia-bles. The removal (expressed as log units) ofthe microbiological indicators was used as thedependent variable in ANCOVA tests; the pres-ence of plants, depth, type of sand, HL and unitdose as fixed factors, and temperature as covari-ate. Statistical significance was established atp � 0.05.

3. Results and discussion

Mean microbial densities (log CFU/100mLfor bacteria and log PFU/100mL for phages) inthe influent and effluent of the filters and removalefficiencies are shown in Table 4 and Fig. 1.These values are the averages and standard devi-ations of the data obtained during 2 years of mon-itoring for all the operational conditions describedin Table 2.

The concentrations of the indicators of fecalcontamination in the pond effluent (filter inlet)were comprised between 104 and 105 CFU orPFU per 100 mL, except for F-specific bacterio-phages, which registered around 103 PFU/100 mL. This lower concentration was due tothe higher removal of these phages in thepond. The average concentrations of the indica-tors at the filter outlets were in general above1000 CFU/100 mL.

The average removal of all the indicators wasless than 2 log units (Fig. 1). The removal of FCand E. coli were similar, between 0.5 and 2 logunits depending on the filter configuration andoperating conditions. The observed bacterialindicators removal were consistent with theresults obtained with other VFCWs operatingat a similar HL. At HL of 50–130 cm/d, aremoval of 0.7–1.5 orders of magnitude oftotal coliforms and faecal streptococci was

Table 3

Analytical methods for the microbiological parameters

Parameter Method

Fecal coliforms Standard Methods [22]Ref. 9222-D. Filtration method

E. coli ISO 9308-1 [23]Filtration Method

Somatic coliphages ISO 10705-2 [24]E. coli CN ATCC 700078 was used as the host strain

F-specific bacteriophages ISO/CD 10705-1 [25]Salmonella typhimurium NCTC 12484 was used as the host strain

A. Torrens et al. / Desalination 247 (2009) 170–179 173

observed by Arias [26]. Vacca et al. [17] showedthat the number of coliform bacteria was reducedby two orders of magnitude in VFCWs at lowerHL (6 cm/d). Sleytr et al. [16] reported highremoval rates for indicator organisms (4.35 logunits for E. coli, 4.31 for total coliforms and4.80 for enterococci) in a VFCW that treats aprimary effluent. However, the HLs applied inthe latter study were much lower: from 6 cm/d(4 feeding sequences per day) to 24 cm/d (8 feed-ing sequences per day) and the inlet concentra-tions were much higher than those applied inour study.

The removal of somatic coliphages andF-specific bacteriophages ranged from 0.4 to1.5 logs and from 1 to 0.2, respectively. Thelog reduction of viral indicators never reached

the level observed for most of the bacterial indi-cators. Viral indicators (especially F-specificbacteriophages) were more resistant than bacte-rial ones in both planted and non-planted filters.This result confirms the findings by Campos[27] and Folch [18] that the removal mecha-nisms of viral and bacterial indicators duringinfiltration in porous media are diverse.

3.1. Effect of the plants

No significant difference in the removal effi-ciency of planted and non-planted vertical flowbeds was observed ( p > 0.05). The overallremoval efficiency of the filter plots for all themicrobiological indicators was similar for filtersof the same depth, regardless of whether the

Table 4

Concentrations (Log CFU/100 mL or Log PFU/100 mL) of the microbiological parameters (mean values and

standard deviations)

M65 R65 C65 M25 R25 C25

Indicator Inlet Outlet Inlet Outlet Inlet Outlet Inlet Outlet Inlet Outlet Inlet Outlet

Fecalcoliform

4.7(0.9)

3.2(1.2)

4.7(0.8)

3.1(1.2)

4.7(0.9)

3.1(1.4)

4.7(0.9)

4.0(0.9)

4.7(0.8)

4.0(1.0)

4.7(0.9)

4.1(1.0)

E. coli 4.4(0.9)

2.9(1.1)

4.5(0.8)

3.0(1.2)

4.4(0.9)

2.9(1.3)

4.4(0.9)

3.9(1.0)

4.5(0.8)

3.7(0.9)

4.4(0.9)

3.8(1.0)

Somaticcoliphages

4.1(0.8)

2.9(1.0)

4.2(0.9)

2.9(0.9)

4.2(0.8)

3.0(1.2)

4.1(0.8)

3.5(0.8)

4.2(0.9)

3.5(0.9)

4.2(0.8)

3.5(0.9)

F-specificbacteriophages

3.6(0.4)

2.9(0.7)

3.5(0.4)

2.8(0.7)

3.5(0.4)

2.7(0.9)

3.6(0.4)

3.2(0.6)

3.5(0.4)

3.3(0.6)

3.5(0.4)

3.2(0.7)

0.00

0.50

1.00

1.50

2.00

2.50

M65 R65 C65 M25 R25 C25

Lo

g U

nit

s

FC

E coli

Somaticcoliphages

F-specificbacteriophages

Figure 1. Average removal efficiencies (Log Units).

174 A. Torrens et al. / Desalination 247 (2009) 170–179

plots were planted or not (Fig. 1). These resultsindicate that the presence of macrophytes is ofminor importance for the removal of microor-ganisms in vertical filters intermittently dosed,which is in concordance with the findings ofSteytr et al. [16] and Vacca et al. [17]. The tracertests showed that planted and non-planted filterspresented the same flow patterns, with similarmean HRT and DTD [28]. Moreover, the oxy-genation conditions in planted and non-plantedfilters were similar. In intermittent sand filtra-tion oxygen transfer into the media is mainlyachieved by convection due to batch loadingand diffusion processes from the air. Eventhough the oxygen concentration transportedby Phragmites could be important in HFCWsthe quantity of oxygen provided by this plantin relation with the other sources of oxygen isnegligible in vertical filters intermittently dosed.

3.2. Effect of filter depth and sand type

The filters with a depth of 65 cm presentedsignificant ( p < 0.05) higher removal of bacte-rial and viral indicators than those of 25 cm(Fig. 1). As expected, tracer tests showed thatwater flow through the filters with a depth of25 cm was faster than through those of 65 cm.It is remarkable that a depth increase of only40 cm resulted in an augmentation of the meanresidence times of 3 to 5 hours. The dose appliedin every feeding sequence allowed saturating the15 cm top layer of the filters. Therefore, the exis-tence of a capillary fringe being taken intoaccount, water pulses infiltrated trough saturatedor nearly saturated 25 cm deep sand beds. Watersaturation assured a maximum hydraulic con-ductivity and high water velocities. The samedose was too small for the water content beingclose to saturation throughout the 65 cm deep fil-ters. As water conductivity decreases very rap-idly with the water content, water transferredmore slowly in the low half part of 65 cm deepfilters. Hence, the short HRTs of the shallow

beds (less than 3 hours in general) may notallow nociceable disinfection.

No significant difference (p > 0.05) wasfound between the two kinds of sand compared(crushed and river) for any microbial indicator(Fig. 1). These results indicate that if the particlesize distribution is suitable, the use of alluvialmaterial is not indispensable. Sand from quarriesmay also be used after crushing and sieving.

3.3. Effect of operational parameters

High HLs resulted in a reduction in disinfec-tion capacity. Figure 2 presents the average logremoval of the microbiological indicators forHL around 20 cm/d, 40 cm/d and 80 cm/d. Theincrease of the HL significantly decreased theremoval rates for FC, E. coli and somaticcoliphage indicators ( p < 0.05). Tracer testsshowed that the detention time decreased withincreased HL. Lower purification, as a result ofincreased dosing rate, has also been observed inseveral experiments with intermittent infiltrationin porous media [7,29]. However, the removalof F-specific bacteriophages was not significantlyaffected by the HL ( p > 0.05). The log removalsof this viral indicator were always lower than 1Ulog irrespective of the HL applied.

The results obtained with the fractionation(f = 16, f = 32) of the total HL (80 cm/day)showed that the high fractionation (lower waterheight per unit dose) significantly increased theremoval efficiency FC, E. coli and somatic coli-phage (p < 0.05) (Fig. 3). Previous studies in infil-tration–percolation systems also demonstratedthat the higher the f value, the greater the removalof faecal indicators [7,18]. However, the removalof F-specific bacteriophages was not significantlyaffected by the fractionation ( p > 0.05).

The DTD was strongly influenced by thefractionation of the daily HL (Fig. 4), asobserved by Brissaud et al. [7] and Molle et al.[30]. For the same HL (75 cm/d), when f waslower (higher water height for unit dose) a

A. Torrens et al. / Desalination 247 (2009) 170–179 175

considerable amount of water applied spent onlya very short time in the bed. Tracer breakthroughoccurred for less than 1 h when f = 15. A highdosing rate increases the water velocity throughlarger pores and reduces the exchange betweenmobile and less mobile water [21]. Hence, by

decreasing the batch volume, retention time isincreased and there is greater exchange of theless mobile fraction of the pore water. Thisaffords closer and longer contact between mediaand microbial indicators and results in enhancedbacterial adsorption and higher purification.

Fecal coliform E Coli

Somatic coliphages F-specific bacteriophages

Ulo

g

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

Ulo

g

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

Ulo

g

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

Ulo

g

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

HL 80 cm/d f = 32

HL 80 cm/d f = 16

HL 80 cm/d f = 32

HL 80 cm/d f = 16

HL 80 cm/d f = 32

HL 80 cm/d f = 16

HL 80 cm/d f = 32

HL 80 cm/d f = 16

Figure 3. Removal of microbiological indicators for a range of batch frequencies (f).

Fecal coliform

Ulo

g

E Coli

Somatic coliphages F-specific bacteriophages

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

Ulo

g

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

Ulo

g

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

Ulo

g

M650.0

1.0

2.0

3.0

R65 C65 M25 R25 C25

HL 20 cm/d

HL 40 cm/d

HL 80 cm/d

HL 20 cm/d

HL 40 cm/d

HL 80 cm/d

HL 20 cm/d

HL 40 cm/d

HL 80 cm/d

HL 20 cm/d

HL 40 cm/d

HL 80 cm/d

Figure 2. Removal of microbiological indicators for a range of HLs.

176 A. Torrens et al. / Desalination 247 (2009) 170–179

If an increased HL is required, a higher num-ber of doses would be preferable to an increasein the volume of the batch for the removal ofmicroorganisms. However, the number of timesthe filter is fed per day must be limited andadapted in each case (sand granulometry, depthof the bed, HL, . . .) to avoid saturated flowregimes. Moreover, excessive fractionation candiminish hydraulic conductivity and reduce infil-tration velocity and oxygen supply, therebythreatening the sustainability of the process [31].

3.4. Temperature effects

Water temperatures displayed a pronouncedannual cycle, with maximum values of 288C insummer and minimum values of 48C in winter.Temperatures 15 cm below the filters varied fromabout 2–38C in February/March up to 30–328Cin July/August. The effluent concentrations variedgreatly between seasons as a result of changes ininfluent concentrations. In the warmer months,the removal of indicators by the pond behind thefilters was high, thereby providing an influent

with a lower concentration of pathogens. However,a clear effect between removal rates and tempera-tures was not detected. Neither wastewater temper-ature nor filter temperature significantly affectedthe removal of microbiological indicators (p >0.05). This result, which is consistent with thatreported by Ausland et al. [8] and Campos [27],indicates that lower temperatures do not limit theremoval of pathogens in ISFs and VFCWs.

3.5. Retention time and removal

The results indicate that the depth of the filter,the HL and the fractionation of the daily HL sig-nificantly affect microorganism removal rates.All these design and operational variables weredirectly related to HRT. Great depth of filters,low HLs and low volume per batch resulted inhigh wastewater retention time. Higher HRTwould enhance the adsorption of the microorgan-isms, which is one of the main mechanisms ofmicroorganism immobilization in porous media.To study the effect of the HRT on the removalof these organisms, a correlation between thelog removal of FC and the mean HRT was per-formed. In addition, a second correlation betweenthe removal rates and a minimum retention timewas done. Although both parameters had a signif-icant effect on the FC removal (p < 0.05), a bettercorrelation was found using the minimum reten-tion time (Fig. 5). The use of a minimum retentiontime instead of mean retention time is more appro-priate for predicting microorganism removal onthe basis of breakthrough curves in unsaturated

R2�0.79

0

0.5

1

1.5

2

2.5

0 200 400 600

Minimum HRT (minuts)

FC

Ulo

g r

emo

val

0

0.5

1

1.5

2

2.5

0 500 1000

Mean HRT (minuts)

FC

Ulo

g r

emo

val

R2�0.63

Figure 5. FC removal (Ulog) plotted against minimum HRT and mean HRT.

0.00

0.01

0.02

0 100 200 300 400 500

Time (min)

E(t

)

f�30

f�15

Figure 4. DTD curves for f = 15 and f = 30 (HL =75 cm/day).

A. Torrens et al. / Desalination 247 (2009) 170–179 177

filters. This minimum retention time would betterrepresent the fraction of the dose applied thatspends the least time in the vertical filters. Thisfast-moving fraction of the flow would contributeto increased numbers of FC in the effluent, as thisfraction is conducted through the largest pores ofthe filter media where conditions are less favour-able for both straining and adsorption [21].

4. Conclusions

The removal of microbial indicators inVFCWs and ISFs depends mainly on the waterretention time in the filter, which in turn dependson the depth of the filter, the HL and the dosevolume per batch. The presence of plants didnot significantly affect the removal of indicatormicroorganisms in our study, which indicatesthat the presence of Phragmites is of minorimportance for the removal of microorganismsin VFCWs intermittently dosed. No significantdifferences were found between the two kindsof sand tested (crushed and river sand). Lowtemperatures did not limit the removal of indica-tor microorganisms in ISFs and VFCWs. Bacte-rial indicators were removed at a higher rate thanviral ones. Somatic coliphages were removed athigher rates than F-specific bacteriophages.

Acknowledgements

This study was financed by the EuropeanUnion (LIFE02 ENV/F/000303– LILIPUB).

References

[1] J. Perkins and C. Hunter, Removal of enteric bac-teria in a surface flow constructed wetland inYorkshire, England, Water Res., 34(6) (2000),1941–1947.

[2] R. Hench, G.K. Bissonnette, A.J. Sexstone, J.G.Coleman, K. Garbutt and J.G. Skousen, Fate ofphysical, chemical, and microbial contaminantsin domestic wastewater following treatment bysmall constructed wetlands, Water Res., 37(4)(2003), 921–927.

[3] E. Huertas, M. Folch and M. Salgot, Wastewaterreclamation through a combination of naturalsystems (infiltration-percolation and constructedwetlands): A solution for small communities,Wat. Sci. Tech., 55(7) (2007), 143–148.

[4] T.K. Stevik, G. Ausland and J.F. Hanssen, Reten-tion and removal of pathogenic bacteria in waste-water percolating through porous media: Areview, Water Res., 38(6) (2004), 1355–1367.

[5] M. Auset, A. Keller, F. Brissaud and V. Lazarova,Intermittent filtration of bacteria and colloids inporous media, Water Resour. Res., 41(9) (2005),W09408 doi:10.1029/2004WR003611.

[6] H. Wand, G. Vacca, P. Kuschk, M. Kr€uger and M.K€astner, Removal of bacteria by filtration inplanted and non-planted sand columns, WaterRes., 41(1) (2007), 159–167.

[7] F. Brissaud, M. Salgot, A. Bancole, C. Campos andM. Folch, Residence time distribution and disin-fection of secondary effluents by infiltration perco-lation, Wat. Sci. Tech., 40(4–5) (1999), 215–222.

[8] G. Ausland, T.K. Stevik, J.F. Hanssen, J.C. Køhlerand P.D. Jenssen, Intermittent filtration of waste-water-removal of fecal coliforms and fecal strepto-cocci, Water Res., 36(14) (2002), 3507–3516.

[9] J. Garcıa, J. Vivar, M. Aromir and R. Mujeriego,Role of hydraulic retention time and granularmedium in microbial removal in tertiary treatmentreed beds, Water Res., 37(11) (2003), 2645–2653.

[10] H. Brix, Do macrophytes play a role in constructedtreatment wetlands? Wat. Sci. Tech., 35(54),(1997), 11–17.

[11] O. Decamp, A. Warren and R. Sanchez, The role ofciliated protozoa in subsurface flow wetlands andtheir potential as bioindicators, Wat. Sci. Tech.,40(3) (1999), 91–98.

[12] V. Ottova, J. Balcarova and J. Vymazal, Microbialcharacteristics of constructed wetlands, Wat. Sci.Tech., 35(5) (1997), 117–124.

[13] F. Rivera, A. Warren, E. Ramirez, O. Decamp,P. Bonilla, E. Gallegos, A. Calderon and J.T.Sanchez, Removal of pathogens from wastewatersby the root zone method (RZM), Wat. Sci. Tech.,32(3) (1995), 211–218.

[14] O. Decamp and A. Warren, Abundance, biomassand viability of bacteria in wastewaters: Impactof treatment in horizontal subsurface flow con-structed wetlands, Water Res., 35(14) (2001),3496–3501.

[15] C.C. Tanner, J.S. Clayton and M.P. Upsdell, Effectof loading rate and planting on treatment of dairy

178 A. Torrens et al. / Desalination 247 (2009) 170–179

farm wastewaters in constructed wetlands-I.Removal of oxygen demand, suspended solids andfaecal coliforms, Water Res., 29(1) (1995), 17–26.

[16] K. Sleytr, A. Tietz, G. Langergraber and R. Haberl,Investigation of bacterial removal during the filtra-tion process in constructed wetlands, Sci. of theTotal Env., 380(1–3) (2007) 173–180.

[17] G. Vacca, H. Wand, M. Nikolausz, P. Kuschk andM. K€astner, Effect of plants and filter materialson bacteria removal in pilot-scale constructedwetlands, Water Res., 39(7) (2005), 1361–1373.

[18] M. Folch, Tratamiento terciario de aguas resi-duales por infiltracion-percolacion, PhD thesis,Soil Science Laboratory, Faculty of Pharmacy,University of Barcelona, 1999.

[19] H. Brix, Do macrophytes play a role in constructedtreatment wetlands? Wat. Sci. Tech., 35(54)(1997), 11–17.

[20] A. Lienard, H. Guellaf and C. Boutin, Choice ofthe sand for sand filters used for secondary treat-ment of wastewater, Wat. Sci. Tech., 44(2–3)(2001), 189–196.

[21] T.K. Stevik, G. Ausland, J.F. Hanssen and P.D.Jenssen, The influence of physical and chemicalfactors on the transport of E. coli through biologi-cal filters for wastewater purification, Water Res.,33(18) (1999), 3701–3706.

[22] APHA–AWWA–WEF, Standard Methods for theExamination of Water and Wastewater, 20th edn.Washington, DC, 1998.

[23] ISO, Water quality-Detection and Enumeration ofEscherichia coli and coliform bacteria. Part 1:Membrane filtration method. ISO 9308-1, Interna-tional Organization for Standardization, Geneva,Switzerland, 2000.

[24] ISO, Water Quality-Detection and Enumeration ofBacteriophages. Part 2: Enumeration of Somatic

Coliphages. ISO 1075-2, International Organizationfor Standardization, Geneva, Switzerland, 2000.

[25] ISO, Water Quality-Detection and Enumeration ofBacteriophages. Part 1: Enumeration of F-SpecificRNA Bacteriophages. ISO 1075-1, InternationalOrganization for Standardization, Geneva,Switzerland, 1995.

[26] C.A. Arias, A. Cabello, H. Brix and N.H. Johansen,Removal of indicator bacteria from municipalwastewater in an experimental two-stage verticalflow constructed wetland system, Wat. Sci. Tech.,48(5) (2003), 35–41.

[27] M. Campos, Indicadores de contaminacion fecal enla reutilizacion de agua residual regeneradaen suelos. PhD thesis, Soil Science Laboratory, Fac-ulty of Pharmacy, University of Barcelona, 1998.

[28] A. Torrens, P. Molle, C. Boutin and M. Salgot,Upgrading pond effluent with vertical flow con-structed wetlands and intermittent sand filters: Com-parison of performances and hydraulic behaviour.In: Proc. of the 10th IWA International Conferenceon Wetland Systems for Water Pollution Control,Lisbon, Portugal, Vol. 3, 2006, 1191–1203.

[29] S. Smith, G.W. Thomas, R.E. White and D.Ritonga, Transport of Escherichia coli throughintact and disturbed soil columns, J. Environ.Qual., 14(1985), 87–91.

[30] P. Molle, A. Lienard, A. Grasmick and A. Iwema,Effect of reeds and feeding operations on hydraulicbehaviour of vertical flow constructed wetlandsunder hydraulic overloads, Wat. Res., 40(3)(2006), 606–612.

[31] F. Brissaud, M. Salgot, M. Folch, M. Auset, E.Huertas and A. Torrens, Wastewater infiltrationpercolation for water reuse and receiving bodyprotection: thirteen years’ experience in Spain,Wat. Sci. Tech., 55(7) (2007), 227–234.

A. Torrens et al. / Desalination 247 (2009) 170–179 179