ww.sciencedirect.com

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 9

Available online at w

ScienceDirect

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

Effect of drinking water treatment processparameters on biological removal of manganesefrom surface water

Victoria W. Hoyland a,*, William R. Knocke a, Joseph O. Falkinham IIIb,Amy Pruden a, Gargi Singh a

a VT Via Department of Civil and Environmental Engineering, Virginia Polytechnic Institute and State University,

Blacksburg, VA 24061, USAb VT Biological Sciences Department, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA

a r t i c l e i n f o

Article history:

Received 17 May 2014

Received in revised form

30 July 2014

Accepted 5 August 2014

Available online 14 August 2014

Keywords:

Manganese

Biofilters

Manganese-oxidizing bacteria

* Corresponding author. 1901 Innovation DriE-mail address: vlwheaton@gmail.com (V

http://dx.doi.org/10.1016/j.watres.2014.08.0060043-1354/© 2014 Elsevier Ltd. All rights rese

a b s t r a c t

Soluble manganese (Mn) presents a significant treatment challenge to many water utilities,

causing aesthetic and operational concerns.While application of free chlorine to oxidizeMn

prior to filtration can be effective, this is not feasible for surfacewater treatment plants using

ozonation followed by biofiltration because it inhibits biological removal of organics.

Manganese-oxidizing bacteria (MOB) readily oxidize Mn in groundwater treatment applica-

tions, which normally involve pH > 7.0. The purpose of this study was to evaluate the po-

tential for biological Mn removal at the lower pH conditions (6.2e6.3) often employed in

enhanced coagulation to optimize organics removal. Four laboratory-scale biofilters were

operated over apHrange of 6.3e7.3. Thebiofilterswereable to oxidizeMnat apHas lowaspH

6.3 with greater than 98%Mn removal. Removal of simulated organic ozonation by-products

was also greater than 90% in all columns. Stress studies indicated that well-acclimatedMOB

canwithstandvariations inMnconcentration (e.g., 0.1e0.2mg/L), hydraulic loading rate (e.g.,

2e4 gpm/ft2; 1.36 � 10�3e2.72 � 10�3 m/s), and temperature (e.g., 7e22 �C) typically found at

surface water treatment plants at least for relatively short (1e2 days) periods of time.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Manganese (Mn) isoftenpresent indrinkingwatersources inthe

reduced Mn(II) or oxidized Mn(IV) form. Mn(II) is the more sol-

uble form, and its oxidation results in the formation of a dark

brown MnO2 precipitate, causing unsightly black discoloration

of water and scaling of pipes and fixtures (Sly et al. 1990). Mn in

drinking water can cause aesthetic and operational concerns at

ve, Suite 2100, Blacksburg.W. Hoyland).

rved.

levels below the EPA Secondary Maximum Contaminant Level

(SMCL) of 0.05mg/L (Sly et al. 1990) andhas recently been linked

to neurotoxic effects in children (Wasserman et al. 2006;

Bouchard et al. 2010). One method for soluble Mn removal is to

applyfreechlorineto theinfluentdirectlyupstreamofagranular

filter bed. The result is a natural greensand effect (NGE) where

Mn(II) is initially adsorbed onto a MnOx(s) surface on themedia

and then subsequently oxidized by free chlorine present in the

filter-applied water (Knocke et al. 1988).

, VA 24060, USA.

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 932

Removal of organic compounds from drinking water

sources is also a growing concern, especially for surface water

treatment plants. One effective approach is to apply ozonation

in order to enhance the biodegradability of the organic com-

pounds prior to biofiltration (Rittmann et al. 1989). Ozonation

can also result in soluble Mn oxidation, although its effec-

tiveness for Mn removal is often suboptimal (Wilczak et al.

1993).

Application of chlorine across the filter media to achieve

NGE is at odds with biological treatment benefits of the filter.

Ideally, an effective biofiltration strategy that removes both

Mn and organic contaminants is desirable. This could poten-

tially be achieved by employingmanganese-oxidizing bacteria

(MOB), which are able to naturally oxidize Mn(II) to Mn(IV).

Biological removal of Mn has been studied extensively in the

treatment of groundwater and successfully applied at facil-

ities (primarily in Europe) since the 1980s (Mouchet, 1992;

Vandenabeele et al. 1992; Hope and Bott, 2004; Katsoyiannis

and Zouboulis, 2004; Pacini et al. 2005; Stembal et al. 2005; Li

et al. 2006; Burger et al. 2008a; Tekerlekopoulou et al. 2008).

Biological Mn removal has traditionally been thought to be

restricted to pH > 7.0e7.5 (Mouchet, 1992), although biological

Mn removal may be possible at lower pH (6.5) (Burger et al.

2008a, 2008b).

The role of MOB in biofiltration for Mn removal in surface

water treatment situations has been far less researched.

Granger et al. (2014) reported biological Mn removal to be

effective at pH 6.0 with phosphorus enhancement; less

effective Mn removal was noted under elevated pH (9e11)

conditions. Ozonation was not part of the treatment process

stream employed in this study; as such, the organic matter

present in the filter-applied water was not readily biodegrad-

able but, instead, was natural organic matter from a lake

source.

MOB are phylogenetically diverse and include various

species (Tebo et al. 2004) that appear to have evolved the

ability to oxidize Mn independently. Several MOB have been

studied in pure culture, including: Pseudomonas putida strain

GB-1 (Okazaki et al. 1997), Leptothrix discophora strain SS1

(Boogerd and Devrind, 1987; Katsoyiannis and Zouboulis,

2004) and L. discophora strain SP-6 (Hope and Bott, 2004;

Burger et al. 2008b), and Bacillus sp. strain SG-1 (Devrind

et al. 1986; Bargar et al. 2000). In particular, Bacillus sp. strain

SG-1 (Devrind et al. 1986; Bargar et al. 2000) and other Mn-

oxidizing Bacillus (Francis and Tebo, 2002; Cerrato et al. 2010)

are able to oxidize Mn in the dormant spore form. Bacteria in

spore form are known to be resistant to a host of environ-

mental threats, including heat and oxidants (Nicholson et al.

2000).

The overall goal of this study was to evaluate the potential

for simultaneous Mn and organics removal via biofiltration

over a range of pH and loading conditions relevant to surface

water treatment facilities, especially those employing up-

stream coagulation and ozonation. Removal of Mn and or-

ganics was evaluated in four lab-scale anthracite coal

biofiltration columns over a pH range of 6.3e7.3. The resil-

ience of the columns to shifts in Mn concentration, hydraulic

loading rate, and temperature was evaluated. The results of

this study provide insight into the range of conditions

conducive to biological Mn removal in surface water

treatment plants and the corresponding role of microbial

communities in accomplishing Mn removal.

2. Materials and methods

2.1. Laboratory-scale biofilters

Lab-scale glass columns were used to simulate the anthracite

coal layer of a biologically active filter at a surface water

treatment plant. Glass columns with 5 ft (1.52 m) length and

1.5 inch (3.81 cm) inner diameter were constructed with

sampling ports at 4e6 inch (10e15 cm) intervals along the

column length. Anthracite coal media (effective size

0.95e1.05 mm; uniformity coefficient <1.4) were obtained

from a full-scale biofilter at the Newport News Waterworks

Lee Hall Treatment Plant (Newport News, VA). The columns

were inoculated with a mixture of six MOB isolated from four

water treatment facilities and distribution systems (Cerrato

et al. 2010). MnO2(s) deposition capability over a range of pH

(6.0e7.4), rapid growth rate, and representation of available

genera were the primary criteria for isolate selection. The

species classifications of the six MOB isolates were Bacillus

pumilus (MB-2 and MB-3), Lysinbacillus sphaericus (MB-17),

Lysinbacillus fusiformis (MB-22), Pseudomonas aeruginosa (MB-

33), and Brevundimonas nasdae (MB-38). The media were inoc-

ulated with the MOB mixture and distributed into the four

columns to a depth of two feet (0.61m). A six-inch gravel layer

(nominal size of 0.25e0.5 in, 0.64e1.3 cm) supported the

anthracite coal media bed, and the columns were shielded

from light with aluminum foil.

Isolates MB-2, MB-3, MB-17, MB-22, and MB-38 were rein-

oculated after 30 days of operation by recirculating feed water

for four days with 2.5 mg/L Mn(II), as recommended by Hope

and Bott (2004), because only the Pseudomonas isolate MB-33

remained detectable on MOB selective media during start-up.

2.1.1. Feed water systemThe initial feedwater systembefore reinoculation consisted of

tap water from Blacksburg, Christiansburg, VPI Water Au-

thority, dechloraminated using granular activated carbon

(Calgon Centaur 12x40) with an empty bed contact time of

10 min. The average water quality characteristics of the tap

water were as follows: pHe 7.7, alkalinitye 49mg/L as CaCO3,

hardness e 46 mg/L as CaCO3, dissolved organic carbon (DOC)

e 1.2 mg/L, ammonia e 0.73 mg/L, Mn e <0.5 mg/L, nitrate e

0.45mg/L as N, phosphatee 0.24mg/L as P, with N and P levels

sufficient to ensure carbon was the limiting nutrient. The

water temperature wasmaintained above 20 �C and aerated to

achieve 7.5e8.0 ppm dissolved oxygen. The water was pum-

ped into the columns at 2 gpm/ft2 (1.36 � 10�3 m/s) and sup-

plemented with soluble Mn, pH adjustment, and

biodegradable organic matter (BOM). Concentrated sulfuric

acid diluted in distilled water was used for pH adjustment.

Manganous sulfate monohydrate (MnSO4$H2O) was added to

distilled water to achieve a 0.1 mg/L soluble Mn feed to the

columns. The Mn concentration was increased 16 days after

startup to 0.5 mg/L Mn in an attempt to stimulate MOB accli-

mation and was returned to 0.1 mg/L once Mn removal was

established across the filters.

Fig. 1 e Schematic of final filter column setup (after day 50).

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 9 33

The BOM added to the feed was intended to simulate post-

ozonation by-products commonly found in water treatment

plants using ozone disinfection. The carbon mixture found in

Elhadi et al. (2006) was modified to target concentrations of

500 mg/L oxalate, 400 mg/L formate, 50 mg/L formaldehyde and

25 mg/L glyoxal. The BOM stock was made by adding sodium

formate, sodium oxalate, 40% glyoxal in water, and 37%

formaldehyde in water with preservatives to high-purity

deionized water from a Nanopure system. The stock

container was shielded from light to minimize degradation of

the carbon compounds.

No Mn removal was observed over the first 30 days of oper-

ation.Microorganisms in theGAC converted the ammonia from

the chloramines into nitrite, so during reinoculation, a break-

point chlorination system was installed prior to the GAC to

oxidize and remove theammoniaasnitrogengas. Thehydraulic

designof thesetupwasalsoreconfiguredduringreinoculationto

relocate BOMaddition directly prior to entry into the columns to

preventBOMdegradationbytubingbiofilm.Fig.1showsthefinal

feedwater system. Further informationanddetails on the initial

feed water system can be found in (Hoyland, 2013).

Table 1 e Summary of laboratory-scale column operating para

Column Initial pH pH after MOB reinoculation Stres

A 6.0 6.3 Mn conc

B 6.3 6.7 None

C 6.7 7.0 Mn conc

Hydrauli

Tempera

D 7.0 7.3 pH

2.1.2. Maintenance and samplingThe columns were backwashed weekly by pumping dechlor-

aminated tap water upflow at approximately 25 gpm/ft2

(1.7 � 10�2 m/s), which corresponded to a 20% bed expansion.

Air in the effluent tube provided an initial air scour before the

backwash water reached the column media. The initial

backwash time was 5 min, which was increased to 7 min

starting Day 93 due to evidence of more extensive head loss in

the columns, most likely from biological growth on the filter

media.

Aqueous samples were withdrawn from sampling ports

using 30 mL syringes with stainless steel needle tips. Influent

samples were collected from the top port above the media

depth and effluent samples were collected from the bottom

sampling port in the gravel layer. Starting on Day 93, samples

for Mn analysis were filtered through 0.45 mm PVDF filters

(Millipore, Billerica, MA) to remove biofilm prior to acidifica-

tion with 2% nitric acid. Media samples were collected by

fluidizing the bed during backwashing, opening the influent

sample port and collecting themedia andwatermixture into a

sterile bottle.

meters.

s test Start day of stress test Duration of stress test

entration 181 24 h

N/A N/A

entration 139 24 h

c loading 157 24 h

ture 226 134 days

233 144 days

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 934

2.2. Stress experiments

Table 1 summarizes the conditions of the four columns (AeD)

and influent stress test performed on each column. Aqueous

samples were collected from the influent, effluent, and profile

sample ports for Mn analysis during the stress tests.

2.2.1. Mn increase stress testThe influent Mn concentration was doubled from 0.1 mg/L to

0.2mg/LMn in Columns A (pH 6.3) and C (pH 7.0) for a period of

approximately 24 h by incrementally increasing Mn concen-

tration in the Mn/pH adjustment stock solution over 2 h,

maintaining the higher concentration, and decreasing the Mn

concentration in the same fashion.

2.2.2. Hydraulic loading rate increase stress testInfluent hydraulic loading rate was doubled from 2 gpm/ft2 to

4 gpm/ft2 (1.36 � 10�3e2.72 � 10�3 m/s) in Column C (pH 7.0)

for approximately 24 h. Batched feed water was used during

incremental increasing and decreasing of the hydraulic

loading rate over 2 h. Experimental conditions for Mn con-

centration, filter-applied pH, and BOM concentration were

maintained during the hydraulic loading rate stress study by

adjustments to stock solution concentrations as necessary.

2.2.3. Temperature decrease stress testA chiller was used to decrease the influent water temperature

to Column C (pH 7.0) approximately 0.5 �C per day for 25 days.

Column C was backwashed with chilled dechloraminated tap

water when head loss reached approximately one foot. The

influent temperature was increased over the course of 20 days

once it reached 7 �C. The Mn removal across Column C was

monitored for an additional 30 days after the column returned

to room temperature (22 �C).

2.2.4. Column D pH adjustmentColumn D (pH 7.3) was expected to achieve greater Mn

removal than the other columns because it was operating

closer to the pH range traditionally associated with effective

biological Mn removal (Mouchet, 1992). After 233 days of

operation at pH 7.3, Mn removal in Column D remained sub-

stantially lower than the other lower pH columns. To explore

the effect of pH on Column D performance an additional

“stress” test was carried out in which the influent pH to Col-

umn D was decreased to pH 6.7 starting Day 233 for 60 days,

before returning to pH 7.3.

2.3. Analytical methods

Aqueous samples were preserved in 2% trace metal grade ni-

tric acid for Mn analysis. Samples above 0.05 mg/L Mn were

analyzed using flame atomic absorption spectroscopy (Fl-

AAS) and below 0.05 mg/L using inductively coupled plasma

mass spectroscopy (ICP-MS). The minimum detection limit

(MDL) was 0.017 mg/L for ICP-MS and 13.4 mg/L for Fl-AAS. pH

was measured using an Accumet pH electrode probe and

Oakton pH 110 series meter. Temperature was measured

using a digital Fisher Scientific traceable thermometer.

Formaldehyde and glyoxal were measured following EPA

Method 556 (Munch et al. 1998). The MDLs for formaldehyde

and glyoxal were determined to be 0.64 and 4.5 mg/L, respec-

tively. Formate and oxalate were measured using a modified

version of the ion chromatography method outlined in

Peldszus et al. (1996, 1998). A 70 mM NaOH eluent was used

because equipment limitations prevented using an eluent

gradient. Samples were preserved with 0.1% v/v chloroform

and analyzed immediately after sampling. The MDLs for

formate and oxalate were determined to be 36 and 6.2 mg/L,

respectively. For the purpose of data analysis, all BOM

componentmeasurements below theMDLwere considered to

be half the MDL value.

Mn-oxidation broth and agar plates were prepared ac-

cording to Stein et al. (2001), adjusted to pH 7.0. The supple-

mented reduced Mn present in the medium allows MOB to be

distinguished from other heterotrophs by their characteristic

brown colony color. Incubation temperature was 30 �C. Cellsuspensions were collected by vortexing media samples at

high speed for 1 min. Spores were isolated following Krieg

(1981) and were plated on the Mn-oxidation agar. Repetitive-

sequence-based polymerase chain reaction (rep-PCR)

(Cangelosi et al. 2004) was used to identify which of the five

reinoculatedMOB isolateswere present in the biofilm 200 days

later.

Denaturing gradient gel electrophoresis (DGGE) of bacterial

16S rRNA genes was applied to select media samples (as

described by Singh et al. (2012)) in order to track the fate of the

isolates in the columns and compare the overall microbial

community composition. Amplified 16S rRNA genes from

select samples from Columns B and D were subjected to 454

pyrosequencing (Research and Testing Laboratories, Lubbock,

Texas).

2.4. Data analysis

The relative intensities of visible DGGE bands were trans-

formed by taking the fourth root and Bray Curtis distance and

were used to construct cluster dendrograms using Primer-E

(Clarke and Gorley, 2006). The significance of clustering was

tested using the Simprof test with significance set at p < 0.05.

Pyrosequencing data was analyzed using Mothur v.1.30.0

using 454 SOP (Schloss et al. 2011). Analysis of similarity test

was done using Simprof test from Clustsig package (Oksanen

et al. 2013) in R version 2.15.3 (R Core Team, 2013). Classical

multidimensional scaling (MDS) analysis and plotting of the

graphs was done using packages MASS (Venables and Ripley,

2002), vegan (Oksanen et al. 2013), and rgl (Adler andMurdoch,

2013) in R (R Core Team, 2013).

3. Results and discussion

3.1. Biofilter column acclimation

3.1.1. Effect of influent pH on Mn removalMn removal was observed within one week of Phase 2 startup

(post-reinoculation) in Columns A (pH 6.3) and B (pH 6.7) and

within three weeks in Column C (pH 7.0) (Fig. 2). Essentially

complete (>98%) Mn removal was achieved in Columns A, B,

and C, supporting prior research results which demonstrated

that MOB can oxidize Mn under pH conditions as low as 6.5

Fig. 2 e Influent and effluent Mn concentration during

startup of the lab-scale column (Column A e Phase 1

pH ¼ 6.0, Phase 2 pH ¼ 6.3; Column B e Phase 1 pH ¼ 6.3,

Phase 2 pH ¼ 6.7; Column C e Phase 1 pH ¼ 6.7, Phase 2

pH ¼ 7.0; Column D e Phase 1 pH ¼ 7.0, Phase 2 pH ¼ 7.3).

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 9 35

(Burger et al. 2008a). Column D (pH 7.3) did not remove an

appreciable amount of Mn. This was contrary to prior reports

that MOB require pH conditions as high as 7.4e7.5 to oxidize

Mn (Mouchet, 1992). Additionally, the columns operating at

lower pH (6.3 and 6.7) began removingMn earlier than those at

higher pH, supporting the findings of Burger et al. (2008b),

where better Mn removal was observed at pH 6.5 than pH 7.5.

The Burger study was carried out in laboratory-scale filter

studies inoculated with L. discophora strain SP-6 and indige-

nous microflora from full-scale treatment facilities where

biological Mn removal was occurring.

Interestingly, despite negligible Mn removal in Column D

over 200 days of acclimation at pH 7.3, removal was readily

stimulated by decreasing the pH to 6.7 (Fig. 3). Mn removal

reached greater than 98% at pH 6.7, but correspondingly

declined again when the pH was returned to 7.3. It is impor-

tant to note that Mn removal improved markedly (in com-

parison to the initial studies at pH 7.3) when the pH was

adjusted up to 7.3 after operating effectively for several days at

pH 6.7. Although Column D Mn removal decreased to 70%

when the pHwas first increased to 7.3, eventually Mn removal

greater than 98% was achieved. This result may indicate that

once acclimated, the MOB were able to tolerate pH 7.3. Alter-

natively, MOB-based formation of MnOx(s) on the media

Fig. 3 e Effect of influent pH on Mn removal in Column D.

surface at pH 6.7 may have allowed physical-chemical Mn

removal processes to contribute to overall Mn removal. Mn

removal in pilot-scale trickling filters has been attributed to

both biological and physical-chemical mechanisms (Gouzinis

et al. 1998). Most of the Mn removal occurred within the first 4

in. (10.2 cm) of the columns (Fig. 4). This is expected since

most of the biological activity occurs at the top of biofilters,

where concentrations of organic carbon and other nutrients

are highest. Physical-chemical Mn oxidation and removal also

occurs primarily at the top of filter media.

The initial period of poor Mn removal (Column D; pH 7.3)

was unexpected based upon the published work of others

(Mouchet, 1992; Katsoyiannis and Zouboulis, 2004; Li et al.

2006; Burger et al. 2008a; Katsoyiannis et al. 2008). A defini-

tive explanation for this difference in performance was not

achieved in the current study. All of the research work cited

above involved the treatment of groundwater sources which

would typically be characterized by low DOC concentrations,

with minimal BDOC present. In contrast, the waters in the

current study had a significant amount of BDOC present.

Future research could investigate whether the presence or

absence of BDOC may impact MOB effectiveness for soluble

Mn removal under alkaline (>pH 7) conditions.

3.1.2. Acclimation timeMn removal was not observed in any of the columns over the

first 50 days (Phase 1) at which time several factors were

adjusted simultaneously in an attempt to stimulate removal:

a) columns were reinoculated with MOB; b) the chloramine

removal method was changed, which changed the form of

nitrogen composition of the influent water; c) Mn influent

concentration was increased from 0.1 mg/L to 0.5 mg/L; and d)

the hydraulic setup was changed to prevent BOM degradation

prior to the columns and provide consistent influent flow.

Because multiple factors were changed simultaneously, it is

not possible to concludewhich, if any, contributed to the rapid

onset of Mn removal in Columns AeC from Day 50 onwards

(Phase 2).

Regardless, acclimation appears to be an important factor

for MOB. Even upon the onset of Mn removal, nearly complete

(98%) removal was not achieved for an additional 50 days.

Fig. 4 e Mn removal profiles across the media depth of the

lab-scale columns once 98% Mn removal was established

in Columns A, B, and C. Profile samples collected on Day

220.

Fig. 5 e Mn removal profile across Column C filter media

depth during influent Mn concentration increase stress

study (influent pH of 7.0).

Fig. 6 e Effect of temperature on Mn removal in Column C

(influent pH of 7.0). Time is displayed as days since

temperature study began. (Day 0 is 226 days after Phase 1

startup).

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 936

Mouchet (1992) documented a similar slow improvement in

Mn removal with longer acclimation times, with the time

between initial Mn removal and near complete removal

requiring approximately 35 days. Additionally, Burger et al.

(2008b) found that a lab-scale groundwater MOB biofilter

required approximately 45 days for acclimation.

The increased Mn concentration during Phase 2 startup of

the lab-scale columns is one factor that may have aided the

eventual stimulation of Mn removal. As previously

mentioned, most studies on the use of MOB in biofiltration

have involved groundwater systems which generally have

higher influent Mn concentration than surface water treat-

ment plants (Mouchet, 1992; Vandenabeele et al. 1992; Hope

and Bott, 2004; Katsoyiannis and Zouboulis, 2004; Li et al.

2006; Burger et al. 2008a). Hope and Bott (2004) recom-

mended an initial Mn concentration of 2.5 mg/L in a recircu-

lation setup for fastest acclimation. The lab-scale columns

were operated in this configuration during reinoculation in

Phase 2 for the first week of operation before starting normal

flow at 0.5 mg/L Mn. This may have contributed to the Mn

removal increase shortly after Phase 2 reinoculation.

3.1.3. Biodegradable organic matter removalBOMwas readily removed across all columns, ranging from 90

to 98% removal for the four compounds (formate, oxalate,

glyoxal, formaldehyde). As BOM removal is the primary pur-

pose of biofiltration, it was an important finding that MOB

inoculation did not adversely affect this process. Similar to

Mn, most (>80%) of the BOM removal occurred in the first 4 in.

(10.2 cm) of the column, as shown in Fig. S-1 in the supple-

mental information.

3.2. Stress experiments

3.2.1. Mn concentrationEffluent Mn concentration did not appreciably increase when

influent Mn concentration doubled from 0.1 to 0.2 mg/L over a

24 h period in Column C (pH 7.0), indicating that acclimated

MOB communities are resilient to short-term variations in Mn

concentration. Influent shifts are common in surface water

treatment plants due to changes in upstream treatment pro-

cesses. Burger et al. (2008b) measured similar results in

acclimated lab-scale groundwater MOB biofilters. The Mn

removal profiles did shift slightly towards higher Mn levels

persisting further down the media depth (Fig. 5), indicating

there may be long-term effects on Mn removal if elevated Mn

levels had persisted. Seasonal long-term variations in Mn

concentration are common at surface water treatment plants;

however, these changes are typically gradual and can be

addressed by changes in upstream treatment processes and/

or MOB acclimation to the higher influent Mn level. Similar

results were observed in aMn increase stress study in Column

A conducted at pH 6.3 (see Fig. S-2).

3.2.2. Hydraulic loading rateHydraulic loading rate may also vary to some degree with

demand in drinking water treatment plants. The results of the

hydraulic loading rate increase stress study of Column C (pH

7.0) were similar to that of the Mn increase stress study. Mn

breakthrough to the filter effluent did not occur during the 24-

h stress period, but the Mn removal profile did shift slightly

over time (Fig. S-3). Stembal et al. (2005) found similar shifting

of Mn removal profiles in groundwater biofilters at increasing

hydraulic loading rates; however, effluent Mn concentration

remained unchanged when the hydraulic loading rate was

doubled.

3.2.3. TemperatureThe lab-scale column temperature decrease study indicated

that well-acclimated MOB can withstand temperatures com-

mon in a mild winter condition (Fig. 6). Mn removal decreased

as influent temperature decreased in Column C (pH 7.0) and

increased once the water temperature began to increase,

eventually returning to greater than 98% Mn removal. Soluble

Mn in surface water sources is typically a summer and fall

concern, when water temperatures would normally be

warmer. As such, biological Mn removal at colder tempera-

tures is not generally necessary, but the ability for MOB to

survive winter conditions on the filter media and readily

removeMn once temperatures increase is critical for full-scale

applications. These results indicate the MOB were not only

able to survive and recover from colder winter temperatures,

they were also able to maintain some oxidative activity

throughout the colder temperature conditions.

Fig. 7 e A) Cluster analysis of denaturing gradient gel

electrophoresis (DGGE) profiles of bacterials 16S rRNA

genes in the four columns. Analysis of D columns is before

and after experiment testing effect of temporarily lowering

the pH. Black lines separate significantly different clusters.

B) Multi-dimensional scaling analysis of pyrosequencing of

bacterial 16S rRNA genes in Column B versus Column D,

before and after experiment testing effect of temporarily

lowering the pH.

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 9 37

3.3. Microbial composition

MOB have been reported to comprise a large percentage

(25e33%) of the total microbial population in mature

groundwater treatment biofilters (Vandenabeele et al. 1992).

The percentage of MOB in mature surface water treatment

biofilters may be much smaller. The percentage MOB with

respect to total heterotrophic plate count (HPC) in the lab-

scale columns was determined using agar plating methods

on Days 104 and 118 when Columns B (pH 6.7) and C (pH 7.0)

were removing greater than 98%Mn, Column Awas removing

around 70% Mn, and Column D was not appreciably removing

Mn. MOB accounted for a very small percentage of total HPC

(3.6% in Column A, 5.7% in Column B, 1.3% in Column C, and

2.1% in Column D); however, all isolated spore-forming col-

onies from each column were identified as MOB.

The rep-PCR analysis indicated that the majority of the

MOB colonies and spores present in the media biofilm corre-

sponded to B. pumilus, which corresponded to strain MB-3

originally isolated by Cerrato et al. (2010) from the Newport

News Lee Hall water treatment plant where the lab-scale

media was obtained (see Figs. S-4 and S-5 and Table S-1).

Bacilli may require stress to induce Mn oxidation as these

species are known to oxidize Mn only in their spore form

(Devrind et al. 1986; Bargar et al. 2000), which could also be a

reason for the extended acclimation time. The MB-3 growth

curves (see Fig. S-6) may provide further insight as to why

ColumnDwas unable to initially remove substantial Mn at pH

7.3. The lag phase of MB-3 is much longer at pH 7.4 than at pH

6.0 or 6.5. MB-3 may grow and oxidize Mn differently at higher

pH than lower pH.

The DGGE cluster analysis results indicated that the mi-

crobial communities present on Column D media before the

pH change (Day 183, 20%Mn removal) and after the pH change

(Day 260, 65% Mn removal) were not statistically different

(p < 0.05, Simprof test). Further, both Column D microbial

communities were different from samples collected (on Day

183) from Columns A, B, and C (p < 0.05 Simprof test.), which

were not significantly different from each other (p < 0.05

Simprof test.). These results suggest that the higher pH in

Column D selected for a distinct microbial community and

that the microbial community directly impacts Mn removal.

Notably, the microbes in Column D exhibited distinct Mn

removal behavior relative to the other columns. Nonetheless,

they did appear capable of acclimating when the pH condi-

tions were temporarily decreased. Further information on the

DGGE results can be found in Tables S-2 and S-3 and Fig. S-7.

Pyrosequencing was applied to compare the bacterial com-

munities in Columns B and D, which exhibited contrasting

performance and harbored distinct bacterial communities

based onDGGE results. Pyrosequencing results suggest that the

bacterial community inColumnDclustered separatelyandwas

significantly different from the bacterial community in column

B (p<0.05, Simprof test)during theiroperation inPhase2 (Fig. 7).

Further, the bacterial community composition of ColumnDdid

not shift significantly during the course of the pH change and

subsequent improvedMn removal from ColumnD (Fig. 7). This

suggests that improvedMn removal in ColumnD following the

pHadjustment to 6.7 and return topH7.3wasnot the result of a

shifting bacterial community but, rather, acclimation of the

existing MOB, such as triggering of spore formation.

4. Conclusion

Laboratory-scale studies demonstrated that MOB have po-

tential for soluble Mn control in surface water treatment

plants. Insight into the ability of MOB to oxidize Mn during

surface water treatment plant over a range of conditions has

been gained. The following conclusions can be drawn from

the results of these experiments:

� MOB can oxidize and achieve >98% removal of soluble Mn

in a biofilter as low as pH 6.3, at a condition which surface

water treatment plants often operate.

� Biological Mn removal can be achieved without adversely

affecting BOM removal (>90% removal of BOM compo-

nents), allowing biofilters to achieve their primary

objective.

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 938

� Once acclimated, MOB can tolerate short-term (e.g. 1e2

days) changes in influent characteristics commonly

occurring at surface water treatment plants, including

influent increases in Mn concentration (from 0.1 to 0.2 mg/

L) and hydraulic loading rate (from 2 gpm/ft2 to 4 gpm/ft2

(1.36 � 10�3e2.72 � 10�3 m/s)) without Mn breakthrough

occurring (>98% Mn removal). Long-term effects of these

changes may be more pronounced.

� MOB can withstand and recover from an influent water

temperature decrease from 22 �C to 7 �C, simulating a mild

winter.

� Pyrosequencing and DGGE indicated that the bacterial

community in Column D was significantly different from

those of Columns A, B and C. This difference in bacterial

community in Column Dmay have contributed to the poor

Mn removal noted initially at pH 7.3. Also, the pH adjust-

ment between 6.7 and 7.3 for Column D did not result in a

significant shift in bacterial community. This suggests that

different bacterial communities may display differing Mn

removal behavior with varying treatment conditions.

� Overall, acclimation and stress events may be more

important factors for successfully achieving Mn removal

by microbial means than other studies on the application

of biological Mn removal in drinking water treatment have

indicated.

Acknowledgments

The authors would like to thank the Via Department of Civil

and Environmental Engineering and the Biological Sciences

Department at Virginia Tech for providing support for this

work.

Appendix A. Supplementary data

Supplementary data related to this article can be found at

http://dx.doi.org/10.1016/j.watres.2014.08.006.

r e f e r e n c e s

Adler, D., Murdoch, D., 2013. rgl: 3D Visualization Device System(OpenGL). R Package Version 0.93.932. http://CRAN.R-project.org/package¼rglgar.

Bargar, J.R., Tebo, B.M., Villinski, J.E., 2000. In situ characterizationof Mn(II) oxidation by spores of the marine Bacillus sp. strainSG-1. Geochim. Cosmochim. Acta 64 (16), 2775e2778.

Boogerd, F.C., Devrind, J.P.M., 1987. Manganese oxidation byLeptothrix discophora. J. Bacteriol. 169 (2), 489e494.

Bouchard, M.F., Sauv�e, S., Barbeau, B., Legrand, M., Brodeur, M.-�E.,Bouffard, T., Limoges, E., Bellinger, D.C., Mergler, D., 2010.Intellectual impairment in school-age children exposed tomanganese from drinking water. Environ. Health Perspect. 119(1), 138e143.

Burger, M.S., Krentz, C.A., Mercer, S.S., Gagnon, G.A., 2008a.Manganese removal and occurrence of manganese oxidizingbacteria in full-scale biofilters. J. Water Supply: Res. Technol. e

AQUA 57 (5), 351e359.Burger, M.S., Mercer, S.S., Shupe, G.D., Gagnon, G.A., 2008b.

Manganese removal during bench-scale biofiltration. WaterRes. 42 (19), 4733e4742.

Cangelosi, G.A., Freeman, R.J., Lewis, K.N., Livingston-Rosanoff, D., Shah, K.S., Milan, S.J., Goldberg, S.V., 2004.Evaluation of a high-throughput repetitive-sequence-basedPCR system for DNA fingerprinting of Mycobacteriumtuberculosis and Mycobacterium avium complex strains. J. Clin.Microbiol. 42 (6), 2685e2693.

Cerrato, J.M., Falkinham, J.O., Dietrich, A.M., Knocke, W.R.,McKinney, C.W., Pruden, A., 2010. Manganese-oxidizing and-reducing microorganisms isolated from biofilms inchlorinated drinking water systems. Water Res. 44 (13),3935e3945.

Clarke, K., Gorley, R., 2006. PRIMER v6: User Manual/Tutorial.PRIMER-E, Plymouth.

Devrind, J.P.M., Devrinddejong, E.W., Devoogt, J.W.H.,Westbroek, P., Boogerd, F.C., Rosson, R.A., 1986. Manganeseoxidation by spores and spore coats of a marine Bacillusspecies. Appl. Environ. Microbiol. 52 (5), 1096e1100.

Elhadi, S.L.N., Huck, P.M., Slawson, R.M., 2006. Factors affectingthe removal of geosmin and MIB in drinking water biofilters. J.Am. Water Works Assoc. 98 (8), 108e119.

Francis, C.A., Tebo, B.M., 2002. Enzymatic manganese(II)oxidation by metabolically dormant spores of diverse bacillusspecies. Appl. Environ. Microbiol. 68 (2), 874e880.

Gouzinis, A., Kosmidis, N., Vayenas, D.V., Lyberatos, G., 1998.Removal of Mn and simultaneous removal of NH3, Fe and Mnfrom potable water using a trickling filter. Water Res. Oxf. 32(8), 2442e2450.

Granger, H.C., Stoddart, A.K., Gagnon, G.A., 2014. Directbiofiltration for manganese removal from surface water. J.Environ. Eng. 140 (4).

Hope, C.K., Bott, T.R., 2004. Laboratory modelling of manganesebiofiltration using biofilms of Leptothrix discophora. WaterRes. 38 (7), 1853e1861.

Hoyland, V., 2013. Evaluating the Use of Manganese-oxidizingBacteria in Surface Water Treatment Plants. Virginia Tech,MS.

Katsoyiannis, I.A., Zikoudi, A., Hug, S.J., 2008. Arsenic removalfrom groundwaters containing iron, ammonium, manganeseand phosphate: a case study from a treatment unit inNorthern Greece. Desalination 224, 330e339.

Katsoyiannis, I.A., Zouboulis, A.I., 2004. Biological treatment ofMn(II) and Fe(II) containing groundwater: kineticconsiderations and product characterization. Water Res. 38(Compendex), 1922e1932.

Knocke, W.R., Hamon, J.R., Thompson, C.P., 1988. Solublemanganese removal on oxide-coated filter media. J. Am.Water Works Assoc. 80 (12), 65e70.

Krieg, N.R., 1981. Enrichment and isolation. In: Gerhardt, P.,Murray, R.G.E., Costilow, R.N., et al. (Eds.), Manual of Methodsfor General Bacteriology. American Society of Microbiology,Washington, DC, pp. 114e115.

Li, D., Zhang, J., Wang, H., Chen, L., Wang, H., 2006. Application ofbiological process to treat the groundwater with highconcentration of iron and manganese. J. Water Supply: Res.Technol. e AQUA 55 (5), 313e320.

Mouchet, P., 1992. From conventional to biological removal of ironand manganese in France. J. Am. Water Works Assoc. 84 (4),158e167.

Munch, J.W., Munch, D.W., Winslow, S.D., Wendelken, S.C.,Pepich, B.V., 1998. EPA Method 556: Determination of CarbonylCompounds in Drinking Water byPentafluorobenzylhydroxylamine Derivatization and CapillaryGas Chromotography with Electron Capture Detection. U.E.P.Agency, Cincinnati, OH.

wat e r r e s e a r c h 6 6 ( 2 0 1 4 ) 3 1e3 9 39

Nicholson, W.L., Munakata, N., Horneck, G., Melosh, H.J.,Setlow, P., 2000. Resistance of Bacillus endospores to extremeterrestrial and extraterrestrial environments. Microbiol. Mol.biology Rev.: MMBR 64 (3), 548.

Okazaki, M., Sugita, T., Shimizu, M., Ohode, Y., Iwamoto, K., deVrind-de Jong, E.W., de Vrind, J.P., Corstjens, P.L., 1997. Partialpurification and characterization of manganese-oxidizingfactors of Pseudomonas fluorescens GB-1. Appl. Environ.Microbiol. 63 (12), 4793e4799.

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R.,O'Hara, R.B., Simpson, G.L., Solymos, P., Stevens, M.H.H.,Wagner, H., 2013. Vegan: Community Ecology Package. R.p.v.2.0e6. http://CRAN.R-project.org/package¼vegan.

Pacini, V.A., Ingallinella, A.M., Sanguinetti, G., 2005. Removal ofiron and manganese using biological roughing up flowfiltration technology. Water Res. 39 (18), 4463e4475.

Peldszus, S., Huck, P.M., Andrews, S.A., 1996. Determination ofshort-chain aliphatic, oxo- and hydroxy-acids in drinkingwater at low microgram per liter concentrations. J.Chromatogr. 723 (1), 27e34.

Peldszus, S., Huck, P.M., Andrews, S.A., 1998. Quantitativedetermination of oxalate and other organic acids in drinkingwater at low microgram/1 concentrations. J. Chromatogr. 793(1), 198e203.

Rittmann, B., Huck, P., Bouwer, E., 1989. Biological treatment ofpublic water supplies. Critical Rev. Environ. Sci. Technol. 19(2), 119e184.

Schloss, P.D., Gevers, D., Westcott, S.L., 2011. Reducing the effectsof PCR amplification and sequencing artifacts on 16S rRNA-based studies. PLoS One 6, e27310.

Singh, G., Pruden, A., Widdowson, M.A., 2012. Influence ofpetroleum deposit geometry on local gradient of electronacceptors and microbial catabolic potential. Environ. Sci.Technol. 46 (11), 5782e5788.

Sly, L.I., Hodgekinson, M.C., Arunpairojana, V., 1990. Depositionof manganese in a drinking water distribution system. Appl.Environ. Microbiol. 56 (3), 628e639.

Stein, L.Y., La Duc, M.T., Grundl, T.J., Nealson, K.H., 2001. Bacterialand archaeal populations associated with freshwaterferromanganous micronodules and sediments. Environ.Microbiol. 3 (1), 10e18.

Stembal, T., Markic, M., Ribicic, N., Briski, F., Sipos, L., 2005.Removal of ammonia, iron and manganese fromgroundwaters of northern Croatia e pilot plant studies.Process Biochem. 40 (1), 327e335.

Team, R.C., 2013. R: a Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna,Austria.

Tebo, B.M., Bargar, J.R., Clement, B.G., Dick, G.J., Murray, K.J.,Parker, D., Verity, R., Webb, S.M., 2004. Biogenic manganeseoxides: properties and mechanisms of formation. Annu. Rev.Earth Planet. Sci. 32 (1), 287e328.

Tekerlekopoulou, A.G., Vasiliadou, I.A., Vayenas, D.V., 2008.Biological manganese removal from potable water usingtrickling filters. Biochem. Eng. J. 38 (3), 292e301.

Vandenabeele, J., Debeer, D., Germonpre, R., Verstraete, W., 1992.Manganese oxidation by microbial consortia from sand filters.Microb. Ecol. 24 (1), 91e108.

Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics withS. Springer, New York.

Wasserman, G.A., Xinhua, L., Parvez, F., Ahsan, H., Levy, D.,Factor-Litvak, P., Kline, J., van Geen, A., Slavkovich, V.,LoIacono, N.J., Zhongqi, C., Yan, Z., Graziano, J.H., 2006. Watermanganese exposure and children's intellectual function inAraihazar, Bangladesh. Environ. Health Perspect. 114 (1), 124.

Wilczak, A., Knocke, W.R., Hubel, R.E., Aieta, E.M., 1993.Manganese control Issues associated with the ozonation ofsurface waters. J. Am. Water Works Assoc. 85 (11).