Bc

Ja

b

Mc

a

ARRA

KBBMFBW

1

1p

cHtptoa(b

BU

h0

Ecological Engineering 91 (2016) 50–57

Contents lists available at ScienceDirect

Ecological Engineering

jo ur nal home p ag e: www.elsev ier .com/ locate /eco leng

ait and scrape: An approach for assessing biofilm microbialommunities on organic media used for gas-phase biofiltration

ason P. Olivera, Kevin A. Jannib, Jonathan S. Schillinga,c,∗

Department of Bioproducts and Biosystems Engineering, University of Minnesota, Kaufert Laboratory, 2004 Folwell Ave., St. Paul, MN 55108, USADepartment of Bioproducts and Biosystems Engineering, University of Minnesota, Biosystems and Agricultural Engineering, 1390 Eckles Ave., St. Paul,N 55108, USA

Institute on the Environment, University of Minnesota, 325 LES Building, 1954 Buford Ave., St. Paul, MN 55108, USA

r t i c l e i n f o

rticle history:eceived 20 February 2015eceived in revised form 7 January 2016ccepted 9 February 2016

eywords:iofilmiofiltericrobial biomass

ungi

a b s t r a c t

Gas-phase biofilters offer effective pollution control for agricultural effluents, but a better understandingof microbial communities responsible for capture and degradation is needed to improve process con-trol. In this study, we developed a wood bait and optimized microbial biofilm sampling for monitoringmicrobial biomarkers (microbial C, ergosterol, DNA) in a full-scale biofilter. Results demonstrated thattargeting biofilm dynamics required removing the biofilm from the wood substrates prior to biomarkerextraction. We identified a sampling threshold for these biofilms of ≥100 mg for accurate and low vari-ability biomarker measurement, a threshold that can inform analyses in other systems or using otherapproaches (i.e. DNA sequencing). Using this approach in a full-scale biofilter revealed that the fungalcontribution (as ergosterol) to total microbial biomass was greatest in the most desiccation-prone area

iomarkerood

of the biofilter. This observation is in-line with results from previous lab-scale studies and could be due,in part, to connectivity between fungal hyphae growing in biofilms and the wood baits, shown by con-focal microscopy. This work provides a targeted sampling strategy for microbial biofilms in gas-phasebiofiltration, adaptable to other pollution control bioreactors and that can be used to study microbialcommunity dynamics in full-scale systems.

© 2016 Published by Elsevier B.V.

. Introduction

.1. Focused microbial ecology can guide and improve biofiltererformance

Bioreactors are an efficient alternative to traditional physi-al and chemical pollution control technologies (Delhoménie andeitz, 2005). Unlike other technologies, bioreactors harness bac-

erial and fungal microbial communities to capture and degradeollutants. The use of a diverse and specialized microbial consor-ium enables the simultaneous treatment of pollutant mixturesf variable concentrations, giving bioreactors great promise for

queous (Schipper et al., 2010), and gaseous agricultural effluentsNicolai et al., 2008). Despite their efficacy, low cost, and compati-ility to most U.S. (Liu et al., 2014) and European farming systems

∗ Corresponding author at: University of Minnesota, Dept. of Bioproducts andiosystems Engineering, Kaufert Laboratory, 2004 Folwell Ave., St. Paul, MN 55108,SA. Tel.: +1 612 6241761; fax: +1 612 6243005.

E-mail address: schillin@umn.edu (J.S. Schilling).

ttp://dx.doi.org/10.1016/j.ecoleng.2016.02.010925-8574/© 2016 Published by Elsevier B.V.

(Hamon et al., 2012; Ubeda et al., 2013), bioreactor deploymenthas been hampered by unpredictable performance. Developingmore reliable bioreactors for agricultural pollutant treatment willrequire an improved understanding of the microbial dynamics thatunderpin stable performance (Briones and Raskin, 2003; Cabrol andMalhautier, 2011).

For mitigating livestock emissions, biofilters are typically used.Biofilter are bioreactors which use microbial growths and theirextracellular products attached to stationary reactor media sur-faces – biofilms – to capture and degrade pollutants from a passingeffluent (Mudliar et al., 2010). These systems can simultaneouslytreat volatile organic compounds (VOCs), hazardous odors (e.g. H2S,NH3), and greenhouse gas emissions (e.g. CH4), but mitigation ishighly variable with efforts to improve control hampered by inad-equately understood, complex biofilm processes (Chen and Hoff,2009). As an example of this complexity, degradation of odors VOCsby biofilm heterotrophic bacteria is facilitated by their nourish-

ment with nitrite and nitrate, byproducts of ammonia-oxidizingbacteria (Kristiansen et al., 2011). But, it has also been shown thatammonia-oxidizing bacteria are spatially restricted in biofilters dueto inhibition by their own byproducts and by competition with

l Engi

Vbcamu

1b

reGeoi2mct2mw2sl2

sFslcitgbv

1f

dcruuamawilr

bimmsggdw

J.P. Oliver et al. / Ecologica

OC-degrading heterotrophs for oxygen (Juhler et al., 2009). Thus,iofilter function is dictated by the same nutritional cross-feedingonsidered esential to the stability of microbial communities (Sethnd Taga, 2014), suggesting the whole biofilm community may beore important to biofilter performance than the study of individ-

al functional groups.

.2. Fungal biomass in biofilms may significantly influenceiofiltration

The role of fungi in supporting biofilter performance andesilience have been particularly overlooked (Ralebitso-Seniort al., 2012), despite their abundance in organic (Cardenas-onzalez et al., 1999) and synthetic media biofilters (Kristiansent al., 2011; Xue et al., 2013). Several fungal attributes give theserganisms unique potential to affect biofilter performance. Thesenclude: (1) their tolerance to desiccation (Kennes and Veiga,004) and acidification (Devinny and Ramesh, 2005), both com-on stresses to biofilter stability; (2) their specialized oxidative

apacities, in many cases driving cometabolism of aromatic pollu-ants and lignin (Estevez et al., 2005; Jorio et al., 2009; Spigno et al.,003), (3) their filamentous (hyphal) growth, which can extend intoedia pore space to increase the effective surface area interfacingith the effluent (Arriaga and Revah, 2005; Ralebitso-Senior et al.,

012 and van Groenestijn et al., 2001); and (4) their unique hyphalurface properties, which enable improved capture of emissionsike hydrophobic volatiles (Kennes and Veiga, 2004; Rene et al.,013).

In the presence of hydrophobic VOCs, fungal biomass has beenhown to increase in hydrophobicity and surface area (Vergara-ernandez et al., 2006). The fungal: bacterial ratio has also beenhown to correlate with the capture of emissions with low disso-ution (Prenafeta-Boldú et al., 2012). These characteristics of fungiould be harnessed to filter livestock emissions in confined hous-ngs,where high exhaust rates reduce residence time and increasehe role for capture, not oxidation. To date, the importance of fun-al biomass has not been explored in full-scale livestock emissioniofilters and an adaptable biofilm monitoring approach tunable toarious full-scale designs is required to enable such investigations.

.3. Adapting monitoring general microbial biomarker approachor full-scale biofilters

Comprehensive monitoring of biofilter microbial communitiesemands high-throughput, to detect small spatial/temporal biofilmhanges in large, heterogeneous bioreactors; this heterogeneity theesult of bioreactor size, fluctuating emissions loading, and non-niform organic media. Wood mulch is one of the most widelysed media in US livestock production emission biofilters (Chennd Hoff, 2009). It is preferred for its low-cost, wide availability,oisture holding capacity, bulk porosity, and ability to support

diverse microbial community (Schmidt et al., 2004). Althoughood mulch promotes robust biofilm formation, its heterogene-

ty complicating microbial sampling (Cabrol et al., 2010), so thatarge sample numbers are required to ensure abiotic and biofilmepresentativeness.

Furthermore, monitoring strategies must target the surfaceiofilm communities where effluents and biocatalysis are at an

nterface. Unlike biofilms on impervious media, biofilms on woodulch are accompanied by active microbial populations within theedium. Thus, while some members of these communities are

ignificant to biofiltration (i.e. facilitating pollutant sorption, biode-

rading pollutants, supporting/antagonizing growth of biocatalyticroups), the significance of others inside the wood may be lessirect (i.e. wood degrading saprobes). Separation of biofilm andood microbiota would enable independent assessments of the

neering 91 (2016) 50–57 51

impacts of their growths on pollutant biocatalysis and the interac-tions of these distinct populations.

One rapid, low-cost microbial monitoring approach suitablefor high-sample numbers, biofilms, and wood microbiota is thedetection of microbial biomarkers. Like any measure of microbialstructure, biomarkers are not free of caveat and can only minimallyresolve complex microbial consortia. However, microbial biomassmeasures can provide relevant functional information, as has beenshown in many natural systems (Joergensen and Wichern, 2008)and in engineered biofilters, to isolate the role of fungi in capturinghydrophobic emissions (Prenafeta-Boldú et al., 2012). Moreover,rapid biomarker sampling to complement or target other deepermicrobial community assessments has been demonstrated in lab-scale biofilter research (Cabrol et al., 2009; Prenafeta-Boldú et al.,2012).

1.4. Aims

There were three main intentions of this research. Our firstobjective was to design a bait system that would provide a relevantsubstrate for monitoring bacterial and fungal biofilms colonizingbiofilter packing media. Our second objective was to optimize theharvest and sampling strategies for these baits to ensure repeat-able, low variability extraction of biofilm microbial biomarkers. Ourthird objective was to evaluate the approach in a full-scale biofil-ter to verify our ability to track key biofilter microbes in the field.By developing a bait and retrieve method, a more reproduciblemonitoring of biofilms should be achievable than what has beenaccomplished using the typical ‘grab’ sample approach. By sam-pling in various areas of a biofilter, our study was geared to capturebiofilm variability across biofilter packing. Biomarkers (microbialcarbon (C), ergosterol) were chosen to monitor bacterial and fun-gal dynamics in biofilters for their functional significance in othersystems, low-cost, high-throughput, and their potential to targetmore detailed molecular assessments.

2. Materials and methods

2.1. Design of wood biofilm baits

Wood baits were developed to enable uniform biofilm sam-pling in the heterogeneous biofilter media, similar to the “coupon”baits used to monitor biofilms in aqueous systems (e.g. Deineset al., 2010). Baits consisted of 3 birch (Betula papyrifera) wafers(6.5 × 2.0 × 0.5 cm) cut from dried lumber with the largest face tan-gential, strung to a numbered aluminum tag with fishing line (Fig.S1 in the Supplementary material). The wafer dimensions resem-bled the average biofilter media size distribution, and replicatewafers enabled replicate biomass measures.

2.2. Optimizing the harvest and sampling of biofilms

To identify a harvest strategy that targeted biofilms and toidentify required biofilm sample sizes for biomarker assessment,a microcosm was used to generate sample material. The micro-cosm consisted of a gas-tight plastic tub packed with sievedwood mulch media (1 cm mesh) from a biofilter treating swineodor. This was plumbed to a 55 gal swine manure storage whichgenerated a polluted effluent for a lab-scale biofiltration system(Fig. S2 in the Supplementary material). Average inlet emissionswere 23.5 ± 13.5 ppm NH3, 565 ± 660 ppb H2S, 518 ± 53 ppm CO2,

38 ± 26 ppm CH4, and 404 ± 80 ppb N2O (n > 500). Baits were incor-porated into the tub and the contents were mixed and wateredweekly to promote homogenous biofilm growth and to maintainwood moisture content between 40 and 60 wt.% dry (100 ◦C, 48 h).

5 l Engi

cam(pierwaMswbsbum

htasaldiPrtm(

2

flpRmwoifs(ce

t(fmsiwftcdwSaDw

2 J.P. Oliver et al. / Ecologica

To test biofilm targeting, baits were harvested from the micro-osm at 8 mo. and subsampled aseptically into 5 pieces using

sterile chisel. One subsample (0.5 × 2.0 × 0.5 cm) was used foreasuring moisture content, and the remaining 4 subsamples

1.5 × 2.0 × 0.5 cm) were subsequently fractioned in different waysrior to microbial testing. Fractioning strategies to target biofilms

ncluded comparing ground and whole samples with biofilmsither attached or detached from the media. To do so, with theemaining subsamples, the 1st with the biofilm attached was lefthole (designated as attached whole), the 2nd with the biofilm

ttached was aseptically ground in a Midas Rex® Electric Boneill (Medtronic Inc., Minneapolis, MN, USA) with modified blades

harpened to an angle (attached ground), the biofilm of the 3rdas scraped free using a sterile single-edge razor blade (detached

iofilm), leaving a biofilm-free bait (wood whole), and the 4th wasimilarly scraped but the bait was ground (wood ground). Microbialiomarker were measured on each sample fractions (n = 10) andn-colonized bait controls (n = 5). See Fig. S1 in the Supplementaryaterial for a schematic of the sample fractioning.To test biofilm sample size requirements, a large pool of biomass

ad to be generated and homogenized to ensure uniformity. To dohis, >30 g of biofilm was scraped from the microcosm baits into

sterile blender containing 100 mL of sterilized Type I water. Thelurry was mixed for 30 s, transferred to a sterile glass petri dish,nd the added water was evaporated by the sterile stream of air in aaminar hood. Intensive sample homogenization is required for theetection of cells in aggregates, but also liberates microbial C and

ntroduces error in microbial biomass measurement (Jenkinson andowlson, 1976). To minimize this error, active cells were allowed toeabsorb the microbial C liberated by sample mixing by incubatinghe homogenized biofilm 2 d at 22 ◦C. Microbial biomarkers were

easured on biofilm subsamples of 10, 30, 50, 100, 150, and 200 gn = 10).

.3. Field evaluation of the biofilm sampling technique

For the field component of this work, we used an up-flow,at-bed biofilter constructed in 2011 on a swine nursery with deep-it manure storage at the University of Minnesota West Centralesearch and Outreach Center in Morris, MN (Fig. S1 in the Supple-entary material). Non-inoculated, fresh birch (B. papyrifera) chipshich passed a 5 cm mesh were used as biofilter media. We focused

n a portion of the biofilter (L: 3.5 m, W: 1.25 m, H: 0.35 m) treat-ng air from a continuously running pit fan (See Janni et al., 2014or details). For the first year of operation, inlet and outlet air wereemi-continuously sampled using an automated instrument trailerJanni et al., 2014). Subsequently, inlet and outlet emissions wereollected in duplicate into 50 L FlexFoil® bags similarly to Akdenizt al. (2011) (Table S1 in the Supplementary material).

For the field test, baits were placed at 4 locations in the media athe time of biofilter construction: 50 and 300 cm from the fan inletnear and far, respectively), and 5 and 30 cm from the biofilter sur-ace (shallow and deep, respectively) (Fig. S1 in the Supplementary

aterial). After 20 mo., 10 baits were collected from each location,tored in sterile plastic baggies and returned to the laboratory once. Baits were subsampled for moisture content, biofilm material

as harvested aseptically using a sterile razor blade, then pooledor each location into a sterile 15 mL centrifuge tube, and gen-ly mixed. Baits were then sectioned aseptically using a sterilizedhisel. Similar to the sampling method by (Jasalavich et al., 2000), arill fitted with a sterile 2 mm dia. titanium bit was used to collectood shavings from inside the bait, through the sectioned face (Fig.

1 in the Supplementary materials, Fig. 4b). Wood shavings werelso pooled in a sterile 15 mL centrifuge tube and gently mixed.rilling was used over milling to improve targeting of internalood communities and to speed aseptic sampling. Pooled biofilms

neering 91 (2016) 50–57

and drillings were subsampled (n = 6) for microbial biomassdeterminations.

2.4. Microbial biomarkers

Total microbial biomass was determined by chloroform fumi-gation extraction (CFE) similar to the methods of Needelman et al.(2001), with sample moisture content raised above 50% for dry sam-ples to enable maximal C flush (Ross, 1989). Samples were extractedin 0.1 M K2SO4 (12.5 mL dry g−1) with orbital shaking (100 rpm) for1 h, then filtered and stored at −20 ◦C. Extracts were later thawed,diluted 1:20 with Type I H20, and analyzed for non-purgeableorganic C on a Shimadzu (Kyoto, Japan) TOC-VCPH analyzer. Extrac-tant salinity was lowered to 0.1 M K2SO4 as no statistical differencebetween this concentration and the recommended 0.5 M solutionwas observed (Fig. S3 in the Supplementary material) and the lesssaline extractant, in addition to sample dilution, helped minimizeanalyzer error and maintenance.

Fungal biomass was measured chromatographically as totalergosterol similar to the methods of Schilling and Jellison (2005).Free and esterified ergosterol were extracted from stored sam-ples (−20 ◦C in 5 mL methanol) by reflux, saponification with KOHand an additional reflux. Solids were removed by centrifugationand total ergosterol was separated from alkaline methanol by 3pentane phase separations. Extracts were combined, evaporated,re-dissolved in methanol, filtered and measured by high perfor-mance liquid chromatography (HPLC) on an Agilent Technologies(Minnetonka, MN, USA) 1200 series HPLC.

2.5. Confocal microscopy

Confocal microscopy was used to visualize fungal hyphae inthe biofilm and wood bait, similar to the methods of Schillinget al. (2013). Baits were tangentially sectioned (15 �m) with aDamon-IEC Minotome microtome cryostat (International Equip-ment Company, Chattanooga, TN, USA) after tissue freezingmedium (TFMTM) (Triangle Biomedical Sciences, Durham, NC,USA) infiltration and storage at −20 ◦C. Sections were mountedonto a slide, stained in the dark with 10 �g mL−1 wheat germagglutinin–tetramethylrhodamine in 1 × PBS (pH 7.4) (WGA–TMR,excitation 555 nm, emission 580 nm, Invitrogen, Calsbad, CA, USA)for 30 min, then rinsed with 1 × PBS. Stained sections were thenimaged using a Nikon A1si confocal system (Nikon Instruments Inc.,Tokyo, Japan), equipped with a 32-channel PMT spectral detector,mounted on a Nikon Ti2000E inverted fluorescence scope. The 561and 405 laser bands were used to excite and separate fungal andwood tissues. Emissions were collected and NIS Elements imag-ing software (Nikon Instruments Inc., Tokyo, Japan) was used tounmix selected spectra. ImageJ 1.47 (National Institute of Health,Bethesda, Maryland, USA) was used to prepare images for publica-tion.

2.6. Statistical analyses

For the sample fractioning experiment, the data could notbe normalized through transformation due to the highly vari-able ranges in measurement when different fractions were beingcompared. As such, the Kruskal–Wallis test was used to test fordifferences in the biomarker measures of the different samplefractions. Tukey HSD was used post-hoc to test for differencesin biomarker means of the different sample fractions. A separateKruskal–Wallis test and Tukey HSD test was run for each biomarker

response.

For the sample size experiment, normal distribution of datawas verified by Shapiro–Wilk normality testing. One-way ANOVAwas used to test for differences in the biomarker measures of the

J.P. Oliver et al. / Ecological Engineering 91 (2016) 50–57 53

Fig. 1. The effect of sample fractioning on extraction of microbial biomarkers from baits incubated 8 mo. in a biofilter microcosm. For each biomarker (n = 10) tested subsamplesincluded baits with an attached biofilm left whole (attached whole), or milled (attached ground), biofilms (detached biofilm), biofilm-free wood left whole (wood whole), orm n-inca es, ando

ddsrfm

btpiAaw0

3

3

af(fmfKsb

illed (wood ground), and the sum of paired biofilm and wood fractions (sum). Nore significantly different ( = 0.05). Boxplots depict the media, 1st and 3rd quartilutliers.

ifferent sample sizes. Tukey HSD was used post-hoc to test forifferences in biomarker means of the different sample sizes. Aeparate ANOVA and Tukey HSD test was run for each biomarkeresponse. Additionally, coefficient of variation (CV) was calculatedor each biomarker for each sample size to look at the change in

easurement variability as sample size was increased.For the field evaluation, normal distribution of data was verified

y Shapiro–Wilk normality testing. One-way ANOVA was used toest for differences in the biomarker measures of the different sam-le locations. Tukey HSD was used post-hoc to test for differences

n biomarker means of the different sample locations. A separateNOVA and Tukey HSD test was run for each biomarker responsend the ergosterol: total microbial C ratios. All statistical analysesere conducted with R 3.0.1 (GNU Project) using the alpha level of

.05.

. Results

.1. Targeting biofilm communities

The microbial biomass inside wood was partially extractedlong with attached biofilms; therefore, we isolated biofilmsrom wood substrate (detached) by scraping prior to analysesFig. 1, Table S2 in the Supplementary material). Significant dif-erences between sample fractions were found for both total

icrobial biomass, measured as microbial C (p < 0.001), and total

ungal biomass, measured as ergosterol (p < 0.001) based onruskal–Wallis tests. Mean values for attached whole fractionstatistically exceeded extractable biomarker values for detachediofilm fractions for both biomarkers based on post-hoc tests. The

ubated treatments (controls) (n = 5) were also tested. Means with different letters quartile values plus or minus the interquartile range times 1.5. Circles represent

range and standard deviation of attached whole biomarker val-ues for total microbial and fungal biomass were also larger thandetached biofilm fractions. Detached biofilm fractions routinely hadthe smallest biomarker values of all the fractions tested. Similarpatterns (data not shown) were seen for total DNA extracted by amodified CTAB extraction (Song et al., 2014).

Extraction of attached whole fractions overestimated detachedbiofilm microbial biomass, and in some cases better representedmaximum extractable microbial biomass of the sample measuredas attached ground, or approximated by the sum of surface andwood biomarkers. For all mean biomarker values, there was nostatistical difference between attached ground, the sum of detachedbiofilm + wood whole, and the sum of detached biofilm + wood ground.Exhaustiveness of the biofilm scrape was verified by the lack ofdifference between these values, and confirmed by confocal obser-vations (data not shown). For total microbial and fungal biomassmean values, attached whole was also not statistically different fromthe attached ground baits. Variability was generally higher in sam-ples that were ground. Similar patterns were seen with total DNA(data not shown).

3.2. Biofilm sample size requirements

Variability was typically higher for small sample sizes, andgenerally declined for all biomarkers when larger samples wereused. ANOVA tests indicated statistical differences between sample

sizes for both biomarkers (p < 0.001). Post-hoc analyses indicatedmicrobial C and ergosterol mean values generally plateaued assample size increased with accurate biomarker detection requiring≥100 mg of biofilm per extraction (Table S3 in the Supplementary

54 J.P. Oliver et al. / Ecological Engi

Fms

mbbfw

3

gfihtoibmid

Fbs

ig. 2. Regressions of the average coefficient of variation (CV) for dry weight nor-alized total microbial C (CFE) or total ergosterol (ERG) (n = 10) and the biofilm

ample weight.

aterial). Variability also stabilized at threshold sample weightsetween 50 and 150 mg for the biomarkers, with increasingiomass sample weight correlating well with the reduction in CVor both microbial biomarkers (Fig. 2). Patterns with total DNAere more variable but similar (data not shown).

.3. Biofilm fungi in full-scale biofilters

In the pooled biofilm samples, a significantly higher total fun-al: total microbial biomass ratio was measured on baits in thear-shallow location than in the other locations (Fig. 3, Table S4n the Supplementary material). ANOVA tests indicated locationsad significantly different mean ratio values (p < 0.001), but onlyhe far-shallow location was different than other locations basedn post-hoc tests (p ≤ 0.002). The shift in the fungal biomass ration the far-shallow location was not driven by an increase in fungal

iomass. Though statistically significant differences in ergosterolean values across locations were found based on ANOVA test-

ng (p < 0.001), the ergosterol levels of the far-shallow locationid not change significantly from the other locations based on

ig. 3. Boxplots depicting total fungal: total microbial biomass ratios (measured as massiofilm (light grey) and wood substrates (dark grey). On the second axis is the mean samtandard deviation represented by the error bars (n = 6). Means with different letters are

neering 91 (2016) 50–57

post-hoc tests (p ≥ 0.054). Instead, a significant loss in total micro-bial biomass drove the change in the biomass ratio, as post-hoc testsindicated biofilms from the far-shallow location had significantlylower microbial C measures than all other locations (p ≤ 0.008).The greatest microbial growths were measured in the far-deepbiofilm samples where mean values of chloroform labile-C andtotal ergosterol were both elevated. Homogenized samples of wooddrillings from the full-scale, flat-bed biofilter showed no differencein the ratio of total fungal: total microbial biomass using our pro-tocol (p = 0.527) (Fig. 3, Table S4 in the Supplementary material).Microbial biomass was more concentrated in biofilms than in wooddrillings. An inverse relationship between the mean biofilm fungal:total microbial ratio and the associated moisture content of eachsample area was also apparent (Fig. 3, Table S4 in the Supplemen-tary material). A negative correlation (R2 = −0.545) was also foundwhen biofilm ratios were compared to sample moisture content,regardless of the sample location.

In the far-shallow location, biofilms were typically variable withreduced thickness, compared to the thick, more uniform biofilmsfrom locations like far-deep. On the more desiccated samples, typ-ical of the far-shallow location, biofilms were quite variable andwere often observed with scattered growths of fungal aerial hyphae(Fig. 4a and b). Confocal micrographs show connectivity betweenthese aerial tufts of hyphae and the filamentous growth of fungalinside the wood matrix (Fig. 4c).

4. Discussion

4.1. Wood baits are effective, and isolating biofilms improvedmeasurement precision

Wood baits were effective for sampling biofilter microbiota,and the biofilms on these baits were successfully targeted by theirremoval prior to biomarker measurement. Biofilm baits for aqueoussystems, known as “coupons”, have been successfully used sincethe 1990s to monitor microbial biofilm communities in aqueousbioreactor systems (Deines et al., 2010). These field studies havesignificantly improved aqueous bioreactor stability. While thesecoupons are typically made of piping materials, wood baits havebeen used similarly for monitoring biofilms in natural stream sys-

tems (Tank and Webster, 1998). To our knowledge, this is the firstuse of a wood bait to sample microbial biofilms in gas-phase biore-actors. Like the use of coupons in aqueous systems, the use ofbiofilm baits in gas-phase biofilters facilitated precision sampling

normalized total ergosterol to total microbial C) of biofilter sampling locations forple moisture content (wt.% dry) of biofilter sampling locations (black diamonds),

significantly different ( = 0.05).

J.P. Oliver et al. / Ecological Engi

Fig. 4. (a) Photograph of a bait from the far-shallow location after the moisture con-tent subsection was removed. A fungal dominant biofilm can be seen growing on thebait surface. (b) Cross section of the bait showing drill holes used to sample internalmicrobial biomass. Aerial hyphae can be seen extending out from the biofilm andwood degradation is evident below the biofilm. The black box indicates the approx-imate location of the confocal image. (c) An unmixed, stitched confocal micrographowi

om

wSttsefooshteatsef

apbaemscdta2meatrar

f biofilm and bait section showing fungal hyphae (stained with WGA-TMR, excitedith a 561 laser) growing aerially (white arrow) and connected to mycelial growths

n and across wood cells. The dashed line indicates the surface of the bait.

f natural biofilms growing in a complex, heterogeneous environ-ent.Scraping the biofilms free from bait surfaces prior to analysis

as necessary for low variability, targeted microbial assessment.mall discrepancies between the mean values for summed frac-ions (biofilm + wood) and the entire sample (attached ground) werehe likely result of milling artifacts and their impact on mea-urement accuracy. With the exception of the attached groundrgosterol measurement, milling increased biomarker variabilityor both attached biofilm and scraped samples. Effective samplingf wood microbiota and complete biomarker extraction dependsn effective pulverization of the wood matrix. However, this inten-ive processing can also degrade microbial cellular components andas been shown to increase measurement error for all biomarkersested (Cabrol et al., 2010; Jenkinson and Powlson, 1976; Newellt al., 1988). Although sampling microbial biomass in wood willlways require some sample processing, scraping biofilms fromhe bait circumvented issues with milling and resulted here inmall biomarker measurement variability. While not thoroughlyxplored, similar trends were also observed for extraction of DNArom the different sample fractions (data not shown).

Extraction of attached samples unsuccessfully targeted biofilms,nd differences in biomarker extraction capacities complicate com-arative analyses. When attached whole samples were extracted,iomarker measures were more variable and typically did notpproximate detached biofilm fractions. Differences in the ability ofxtraction protocols to dissociate biofilms and penetrate the woodatrix may explain why biomarker measures of unfractionated

amples overestimated the biofilm. This would be an importantonsideration for the extraction of DNA from intact biofilms foreeper probes of the microbial communities. Biofilms are a pro-ective microbial growth strategy (Hall-Stoodley et al., 2004) withn improved resistance to antimicrobial agents (Mah and O’Toole,001) and as we saw in our preliminary molecular work-ups,olecular assessments in biofilter systems have already shown

xtraction of microbial DNA from active biofilms to be problem-tic (Cabrol et al., 2010). With DNA and biomarker extraction

echniques utilizing different hot solvents, fumigants, and cell dis-uption steps, their abilities to overcome biofilm protection andccess wood microbiota is highly unpredictable, justifying biofilmemoval prior to analysis.

neering 91 (2016) 50–57 55

Sample storage further affected biomarker analysis, particu-larly for ergosterol. In an accompanying methods developmentexperiment, extractable ergosterol from attached whole fractionsincreased with length of storage time, eventually reaching lev-els not statistically different from attached ground fractions (Fig.S4 in the Supplementary material). This additional source of vari-ability might explain the unique result of greater measurementvariability for attached whole fractions than for attached ground frac-tions. In the accompanying experiment, where degradation wasstandardized by use of a single fungal strain, milled sample vari-ability was greater than whole sample variability. It is possiblethat the biofilm and bait test material conditions generated herewere more varied, as a consortium inoculum was used. Conse-quently, inconsistent solvent diffusivity through attached wholesamples may have contributed more to measurement variabilitythan milling where a well-mixed, uniform sample enabled uni-form biomarker extraction. Selecting storage time or extractionconditions that would limit over-estimation of biofilm biomarkerswould be difficult with variable biofilm and bait field conditions.Biofilm scraping (facilitated by the use of baits), is therefore a morestandardizable approach for targeting biofilms and is suitable forvarious biomarkers, environmental condition, and avoids the lim-itations encountered when directly sampling bulk, heterogeneousfield biofilter media.

4.2. The effect of biofilm sample weight on the extraction ofmicrobial biomarkers

Sample size relationships indicated a threshold requirementof ≥100 mg of biofilm for stable, low variability measurementof biomarkers. Based on biofilter sampling experiences from thisstudy, approximately 1 mg of wet biomass was harvested on aver-age from each 1 cm2 of wood bait surface. Thus, at least 100 cm2

of surface area was required per microbial measure. This samplerequirement may be particularly helpful when employing molecu-lar approaches, to avoid under-sampling biofilm surface areas.

In our study, adequate sample size was achieved by poolingscrapes from smaller baits, as baits with a 100 cm2 surface areawould have misrepresented actual biofilter media dimensions andairflow exposures. Pooling also ensured sample representativeness,as some collected biofilm growths had wet weights an order ofmagnitude smaller (desiccated and early time collections) or larger(wetted and later collections) than the mean biofilm size. Highvariability of biofilm thickness in response to localized abiotic con-ditions is not uncommon (Deshusses, 1997; Wäsche et al., 2002).Although pooling may mask some microbial populations in soil(Manter et al., 2010) and biofilm microbial communities (Nyvadet al., 2013), no significant difference was seen between micro-bial populations of individual and pooled composite soil samples(Osborne et al., 2011), and bioreactor studies have shown pool-ing reduces sampling bias and variability (McIlroy et al., 2009).With biofilm growths variable within and across systems, we reportour findings on a mass basis to facilitate adaptation and poolingoptimization for other pollution control bioreactor study systems.Once optimized for a new system, the use of baits allows microbialbiomarkers to be normalized more relevantly by surface area.

4.3. The fungal: Total microbial ratio was greatest in biofilmsharvested from drier biofilter media

As could be done for emissions reductions, the optimizedmonitoring approach successfully correlated a full-scale biofilm

community shift across a small spatial scale to abiotic conditions.Specifically, microbial biomass inside the wood was shown toremain stable across sample locations while fungi dominatedthe biofilms in drier areas of the biofilter. The greatest microbial

5 l Engi

botgg(fiwfffsffst

ihshfdhoo2titfd1q

5

ipbwmtfbid

A

SA8Mnawwtsm

6 J.P. Oliver et al. / Ecologica

iomass, fungal and total, was in the far-deep location, an inlet areaf the biofilter less affected by solar and wind drying. More stableemperature and moisture conditions support vigorous microbialrowth, and reported cell densities in the laboratory are usuallyreatest at biofilter inlets where emissions loading is highestCabrol and Malhautier, 2011). The use of baits corroborated thosendings in a field-scale biofilter. When the proportions of biomassere investigated, however, fungi were found to dominate biofilms

rom drier samples, such as those collected in the more exposedar-shallow location. Biomarker data suggested that greater totalungal: total microbial ratios in these biofilms were the result ofignificant losses in total microbial biomass and not additionalungal biomass. It is plausible that bacterial susceptibility andungal tolerance to the rapid and frequent drying drove thesehifts, supporting other lab-scale observations of fungi betterolerating periodic biofilter desiccation (Kennes and Veiga, 2004).

The potential tolerance of fungi to drier media observedn this study may be explained by the fungal production ofydrophobins and to their unique hyphal growth form. Onceecreted, hydrophobins assemble into surface membranes onyphal and spore surfaces (Linder, 2009). Work with a pathogenic

ungal system has shown this protein coating facilitates sporeesiccation tolerance (Klimes and Dobinson, 2006). Perhapsydrophobin assembly on biofilter hyphae also creates an impervi-us barrier to moisture loss, in addition to their known facilitationf hyphal protrusion into the air from aqueous biofilms (Wosten,001). Fungal morphology may also help explain tolerance to biofil-er media desiccation. As shown in our confocal micrograph, theres connectivity between the biofilm and hyphal reserves insidehe wood. This connectivity may allow fungi to conduit waterrom the wood to drying surface growths, in-line with their well-ocumented capacities to translocate water and nutrients (Cairney,992). Additionally, this “harbor” of inoculum may enable fungi touickly redevelop biofilms when media moisture returns.

. Conclusions

Here we have detailed an optimized wood ‘bait and scrape’ mon-toring technique that can be used to explore the patterns andotentials of bacterial and fungal biofilm growths in gas-phaseiofilters. A sampling threshold of ≥100 mg of microbial biofilmas identified for repeatable, low variability biomarker measure-ent. Through field testing, we have evidence of fungal tolerance

o media mositure stress, and suggest further investigation of fungior their role in biofilter dessication tolerance. Overall, this targetediofilm sampling approach enabled full-scale microbial monitor-

ng of biofilters and can be optimized to facilitate more detailedownstream analyses.

cknowledgements

Funding for this project was provided by grants from the Unitedtates Department of Agriculture, National Institute of Food andgriculture, Agriculture and Food Research Initiative (USDA/2010-5112-20520 & USDA/2012-69002-19880) and the USDA NIFAcintire Stennis Project #MIN-12-074 at the University of Min-

esota. The authors wish to gratefully acknowledge Brian Hetchlernd Larry Jacobsen for their biofilter design, installation and fieldork expertise, members of the Schilling laboratory for assistance

ith sample processing, the United States Forest Service Labora-

ory in Grand Rapids, MN for use of laboratory equipment and thetaff of the University of Minnesota Imaging Center for confocalicroscopy assistance.

neering 91 (2016) 50–57

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ecoleng.2016.02.010.

References

Akdeniz, N., Janni, K.A., Salnikov, I.A., 2011. Biofilter performance of pine nuggetsand lava rock as media. Bioresour. Technol. 102 (8), 4974–4980.

Arriaga, S., Revah, S., 2005. Improving hexane removal by enhancing fungal devel-opment in a microbial consortium biofilter. Biotechnol. Bioeng. 90 (1), 107–115.

Briones, A., Raskin, L., 2003. Diversity and dynamics of microbial communities inengineered environments and their implications for process stability. Curr. Opin.Biotechnol. 14 (3), 270–276.

Cabrol, L., Malhautier, L., 2011. Integrating microbial ecology in bioprocess under-standing: the case of gas biofiltration. Appl. Microbiol. Biotechnol. 90 (3),837–849.

Cabrol, L., Malhautier, L., Poly, F., Lepeuple, A.S., Fanlo, J.L., 2009. Shock loading inbiofilters: impact on biodegradation activity distribution and resilience capacity.Water Sci. Technol. 59 (7), 1307–1314.

Cabrol, L., Malhautier, L., Poly, F., Lepeuple, A.S., Fanlo, J.L., 2010. Assessing the biaslinked to DNA recovery from biofiltration woodchips for microbial communityinvestigation by fingerprinting. Appl. Microbiol. Biotechnol. 85 (3), 779–790.

Cairney, J.W.G., 1992. Translocation of solutes in ectomycorrhizal and saprotrophicrhizomorphs. Mycol. Res. 96 (2), 135–141.

Cardenas-Gonzalez, B., Ergas, S.J., Switzenbaum, M.S., Phillibert, N., 1999. Evaluationof full-scale biofilter media performance. Environ. Prog. 18 (3), 205–211.

Chen, L., Hoff, S., 2009. Mitigating odors from agricultural facilities: a review ofliterature concerning biofilters. Appl. Eng. Agric. 25 (5), 751–766.

Deines, P., Sekar, R., Husband, P.S., Boxall, J.B., Osborn, A.M., Biggs, C.A., 2010. Anew coupon design for simultaneous analysis of in situ microbial biofilm for-mation and community structure in drinking water distribution systems. Appl.Microbiol. Biotechnol. 87 (2), 749–756.

Delhoménie, L.-C., Heitz, M., 2005. Biofiltration of air: a review. Crit. Rev. Biotechnol.25, 53–72.

Deshusses, M.A., 1997. Biological waste air treatment in biofilters. Curr. Opin.Biotechnol. 8 (3), 335–339.

Devinny, J.S., Ramesh, J., 2005. A phenomenological review of biofilter models. Chem.Eng. J. 113 (2–3), 187–196.

Estevez, E., Veiga, M.C., Kennes, C., 2005. Biofiltration of waste gases with the fungiExophiala oligosperma and Paecilomyces variotii. Appl. Microbiol. Biotechnol. 67(4), 563–568.

Hall-Stoodley, L., Costerton, J.W., Stoodley, P., 2004. Bacterial biofilms: from thenatural environment to infectious diseases. Nat. Rev. Microbiol. 2 (2), 95–108.

Hamon, L., Andres, Y., Dumont, E., 2012. Aerial pollutants in swine buildings: areview of their characterization and methods to reduce them. Environ. Sci.Technol. 46 (22), 12287–12301.

Janni, K.A., Jacobson, L.D., Hetchler, B.P., Oliver, J.P., Johnston, L.J., 2014. Semi-continuous air sampling versus 24-hour bag samples to evaluate biofilters on aswine nursery in warm weather. Trans. ASABE 57 (5), 1501–1515.

Jasalavich, C.A., Ostrofsky, A., Jellison, J., 2000. Detection and identification ofdecay fungi in spruce wood by restriction fragment length polymorphismanalysis of amplified genes encoding rRNA. Appl. Environ. Microbiol. 66 (11),4725–4734.

Jenkinson, D.S., Powlson, D.S., 1976. The effects of biocidal treatments on metabolismin soil—V. A method for measuring soil biomass. Soil Biol. Biochem. 8 (3),209–213.

Joergensen, R.G., Wichern, F., 2008. Quantitative assessment of the fungal contribu-tion to microbial tissue in soil. Soil Biol. Biochem. 40 (12), 2977–2991.

Jorio, H., Jin, Y.M., Elmrini, H., Nikiema, J., Brzezinski, R., Heitz, M., 2009. Treatmentof VOCs in biofilters inoculated with fungi and microbial consortium. Environ.Technol. 30 (5), 477–485.

Juhler, S., Revsbeck, N.P., Schramm, A., Herrmann, M., Ottosen, L.D.M., Nielsen, L.P.,2009. Distribution and rate of microbial processes in an ammonia-loaded airfilter biofilm. Appl. Environ. Microbiol. 75 (11), 3705–3713.

Kennes, C., Veiga, M.C., 2004. Fungal biocatalysts in the biofiltration of VOC-pollutedair. J. Biotechnol. 113 (1–3), 305–319.

Klimes, A., Dobinson, K.F., 2006. A hydrophobin gene, VDH1, is involved in microscle-rotial development and spore viability in the plant pathogen Verticillium dahliae.Fungal Genet. Biol. 43 (4), 283–294.

Kristiansen, A., Pedersen, K.H., Nielsen, P.H., Nielsen, L.P., Nielsen, J.L., Schramm, A.,2011. Bacterial community structure of a full-scale biofilter treating pig houseexhaust air. Syst. Appl. Microbiol. 34 (5), 344–352.

Linder, M.B., 2009. Hydrophobins: proteins that self assemble at interfaces. Curr.Opin. Colloid Interface Sci. 14 (5), 356–363.

Liu, Z., Powers, W., Mukhtar, S., 2014. A review of practices and technologies for odorcontrol in swine production facilities. Appl. Eng. Agric. 30 (3), 477–492.

Mah, T.F.C., O’Toole, G.A., 2001. Mechanisms of biofilm resistance to antimicrobialagents. Trends Microbiol. 9 (1), 34–39.

Manter, D.K., Weir, T.L., Vivanco, J.M., 2010. Negative effects of sample pooling onPCR-based estimates of soil microbial richness and community structure. Appl.Environ. Microbiol. 76 (7), 2086–2090.

l Engi

M

M

N

N

N

N

O

P

R

R

R

J.P. Oliver et al. / Ecologica

cIlroy, S.J., Porter, K., Seviour, R.J., Tillett, D., 2009. Extracting nucleic acids fromactivated sludge which reflect community population diversity. Antonie VanLeeuwenhoek 96 (4), 593–605.

udliar, S., Giri, B., Padoley, K., Satpute, D., Dixit, R., Bhatt, P., Pandey, R., Juwarkar,A., Vaidya, A., 2010. Bioreactors for treatment of VOCs and odour—a review. J.Environ. Manage. 91, 1039–1054.

eedelman, B.A., Wander, M.M., Shi, G.S., 2001. Organic carbon extraction efficiencyin chloroform fumigated and non-fumigated soils. Soil Sci. Soc. Am. J. 65 (6),1731–1733.

ewell, S.Y., Arsuffi, T.L., Fallon, R.D., 1988. Fundamental procedures for thedetermining of ergosterol content of decaying plant-material by liquid-chromatography. Appl. Environ. Microbiol. 54 (7), 1876–1879.

icolai, R.E., Janni, K.A., Schmidt, D., 2008. Biofiltration—mitigation odor and gasemissions from animal operations. In: Muhlbauer, E., Moody, L., Burns, R. (Eds.),Proceedings of the National Conference on Mitigating Air emissions from AnimalFeeding Operations. Des Moines, Iowa.

yvad, B., Crielaard, W., Mira, A., Takahashi, N., Beighton, D., 2013. Dental cariesfrom a molecular microbiological perspective. Caries Res. 47 (2), 89–102.

sborne, C.A., Zwart, A.B., Broadhurst, L.M., Young, A.G., Richardson, A.E., 2011. Theinfluence of sampling strategies and spatial variation on the detected soil bac-terial communities under three different land-use types. FEMS Microbiol. Ecol.78 (1), 70–79.

renafeta-Boldú, F.X., Guivernau, M., Gallastegui, G., Vinas, M., de Hoog, G.S., Elías,A., 2012. Fungal/bacterial interactions during the biodegradation of TEX hydro-carbons (toluene, ethylbenzene and p-xylene) in gas biofilters operated underxerophilic conditions. FEMS Microbiol. Ecol. 80 (3), 722–734.

alebitso-Senior, T.K., Senior, E., Di Felice, R., Jarvis, K., 2012. Waste gas biofiltration:advances and limitations of current approaches in microbiology. Environ. Sci.Technol. 46 (16), 8542–8573.

ene, E.R., Veiga, M.C., Kennes, C., 2013. Biofilters. In: Kennes, C., Veiga, M.C. (Eds.),

Air Pollution Prevention and Control: Bioreactors and Bioenergy. , first ed. JohnWiley & Sons, West Sussex, UK.

oss, D.J., 1989. Estimation of soil microbial-C by a fumigation-extractionprocedure—influence of soil-moisture content. Soil Biol. Biochem. 21 (6),767–772.

neering 91 (2016) 50–57 57

Schilling, J.S., Duncan, S.M., Presley, G.N., Filley, T.R., Jurgens, J.A., Blanchette, R.A.,2013. Colocalizing incipient reactions in wood degraded by the brown rot fungusPostia placenta. Int. Biodeterior. Biodegrad. 83 (0), 56–62.

Schilling, J.S., Jellison, J., 2005. Oxalate reduction by two brown rot fungi decayingoxalate-amended and non-amended wood. Holzforschung 59, 681–688.

Schipper, L.A., Robertson, W.D., Gold, A.J., Jaynes, D.B., Cameron, S.C., 2010. Denitri-fying bioreactors—an approach for reducing nitrate loads to receiving waters.Ecol. Eng. 36 (11), 1532–1543.

Schmidt, D., Janni, K., Nicolai, R., 2004. Biofilter design information. In: Biosystemsand Agricultural Engineering Update (BAEU-18). University of Minnesota Exten-sion Service, St. Paul, MN.

Seth, E.C., Taga, M.E., 2014. Nutrient cross-feeding in the microbial world. Front.Microbiol. 5 (350), 1–6.

Song, Z., Vail, A., Sadowsky, M.J., Schilling, J.S., 2014. Quantitative PCR for measuringbiomass of decomposer fungi in planta. Fungal Ecol. 7, 39–46.

Spigno, G., Pagella, C., Fumi, M.D., Molteni, R., De Faveri, D.M., 2003. VOCs removalfrom waste gases: gas-phase bioreactor for the abatement of hexane byAspergillus niger. Chem. Eng. Sci. 58 (3–6), 739–746.

Tank, J.L., Webster, J.R., 1998. Interaction of substrate and nutrient availability onwood biofilm processes in streams. Ecol. 79 (6), 2168–2179.

Ubeda, Y., Lopez-Jimenez, P.A., Nicolas, J., Calvet, S., 2013. Strategies to control odoursin livestock facilities: a critical review. Span. J. Agric. Res. 11 (4), 1004–1015.

van Groenestijn, J.W., van Heiningen, W.N.M., Kraakman, N.J.R., 2001. Biofiltersbased on the action of fungi. Water Sci. Technol. 44 (9), 227–232.

Vergara-Fernandez, A., Van Haaren, B., Revah, S., 2006. Phase partition of gaseoushexane and surface hydrophobicity of Fusarium solani when grown in liquid andsolid media with hexanol and hexane. Biotechnol. Lett. 28 (24), 2011–2017.

Wäsche, S., Horn, H., Hempel, D.C., 2002. Influence of growth conditions on biofilmdevelopment and mass transfer at the bulk/biofilm interface. Water Res. 36 (19),4775–4784.

Wosten, H.A.B., 2001. Hydrophobins: multipurpose proteins. Annu. Rev. Microbiol.55, 625–646.

Xue, N.T., Wang, Q.H., Wang, J., Wang, J.H., Sun, X.H., 2013. Odorous compostinggas abatement and microbial community diversity in a biotrickling filter. Int.Biodeterior. Biodegrad. 82, 73–80.