ORIGINAL ARTICLE

Comparison of the Cecal Microbiota of Domesticand Wild Turkeys

Alexandra J Scupham & Toni G. Patton &

Elizabeth Bent & Darrell O. Bayles

Received: 31 July 2007 /Accepted: 14 November 2007 /Published online: 8 January 2008# Springer Science + Business Media, LLC 2007

Abstract The extent to which production methods alterintestinal microbial communities of livestock is currentlyunknown. As the intestinal microbiota may affect animalhealth, nutrition, and food safety, a baseline comparison of thececal communities of domestic and wild turkeys wasperformed. Oligonucleotide fingerprinting of ribosomalRNA (rRNA) genes (OFRG) of 2,990 16S rRNA clonesand dot blot quantification of dominant populations wereused to identify the dominant bacterial taxa. Seventy-threepercent of all the clones belonged to as yet uncultured genera.However, at a higher phylogenetic level, the OFRG librarywas composed of 54% Bacteroidetes clones (52% of thedomestic library clones, 56% of the wild library clones), 30%Firmicutes clones (33% of the domestic library clones, 32%of the wild library clones), 3% Proteobacteria clones (5%domestic, 2% wild), and 3% Deferribacteres clones (4%domestic, 1% wild). Seven percent of the clones wereunidentifiable (6% domestic, 9% wild). Bacteroidetes clones

included the genera Alistipes, Prevotella, Megamonas, andBacteroides. Of the Clostridiales clones, groups IV, IX, andXIV including genera Faecalibacterium, Megasphaera,Phascolarctobacterium, and Papillibacter were predomi-nant. Lactobacillus, Enterococcus, and Streptococcus bacilliwere also identified. β- δ- and γ-proteobacterial generaincluded Acinetobacter, Sutterella, and Escherichia. Defer-ribacteres clones showed high similarity to Mucispirillumschaedleri. Statistical comparison of the domestic and wildturkey clone libraries indicated similar levels of communityrichness and evenness despite the fact that the two librariesshared only 30% of the total clone operational taxonomicunits. Together these results indicate that although high leveltaxonomic community structure is similar, high-densityturkey production causes considerable divergence of thegenera found in the ceca of commercial birds from those oftheir wild counterparts.

Introduction

It has been proposed that the high levels of disease andcolonization by food-borne pathogens in commercially raisedanimals are caused by current production practices. Somedefining factors of modern production include high-densityrearing conditions, strictly controlled diets and treatment ofpoultry and livestock with an array of vaccines, growth-promoting and therapeutic antibiotics, coccidiostats andfungicides, all of which may affect the intestinal ecosystem.Recent work has shown long-term effects on the intestinalmicrobiota after therapeutic antibiotic administration, includ-ing decreased diversity and significant antibiotic resistance[26]. In modern production facilities, the frequency oftherapeutic antibiotic use suggests that not only are intestinalcommunities of the animals likely different from those of

Microb Ecol (2008) 56:322–331DOI 10.1007/s00248-007-9349-4

Electronic supplementary material The online version of this article(doi:10.1007/s00248-007-9349-4) contains supplementary material,which is available to authorized users.

A. J Scupham (*) : T. G. PattonPre-Harvest Food Safety and Enteric Diseases Research Unit,National Animal Disease Center, Agricultural Research Service,US Department of Agriculture,Ames, IA 50010, USAe-mail: alexandra.scupham@ars.usda.gov

E. BentDepartment of Plant Pathology, University of California,Riverside, CA 92521, USA

D. O. BaylesBacterial Diseases of Livestock Research Unit,National Animal Disease Center, Agricultural Research Service,US Department of Agriculture,Ames, IA 50010, USA

animals naïve to antimicrobials, but the bacterial communi-ties of the treated animals may be in constant flux. It ispossible that disruption of the intestinal ecosystem can allowcolonization of that habitat by pathogens as classical ecologyequilibrium theory states that in a stable ecosystem all nichesare filled, thus excluding opportunistic pathogens [33].

Early studies of the intestinal communities of wild fowlcompared to their domesticated counterparts indicatedsignificant differences between the two groups [5, 20, 38].Using microscopy and aerobic culture techniques, thestudies specifically identified numerous spiral organismsin wild birds not present in domestic birds as well as greaternumbers of aerobic microbes present in the domestic birds[20, 38]. Recent publications on competitive exclusion ofthe food-borne pathogens Salmonella and Campylobacterfrom turkeys have found antagonistic bacterial strains in theintestines of commercial animals [9, 44]. Comparison ofLactobacilli from commercial and free-range chickensindicated that not only did different species inhabit thetwo groups, but isolates from the free-range animals werealso more likely to inhibit a variety of pathogens [43]. Thecurrent work examines community differences betweencommercially raised and wild turkeys to determine whetherantagonism studies would benefit from expansion intosamples from wild birds.

Materials and Methods

Animal Husbandry Twenty-six turkeys, 13 from commer-cial flocks and 13 wild birds, were examined in this surveyof cecal bacterial communities (Table 1). Of the 13domestic birds, five from Farm Awere necropsied on-farm,whereas samples from another five from Farm A and threefrom Farm B were obtained at the slaughterhouse. Cecafrom on-farm necropsy were frozen on dry ice immediatelyafter the animals were killed; in the case of abattoirsamples, ceca were frozen approximately 20 min post-kill.These 13 birds had been raised to 18 weeks of ageaccording to industry standards, with a two-stage growingprogram and exposure to growth-promoting and therapeuticantibiotics, coccidiostats, and vaccinations against hemor-rhagic enteritis and Newcastle’s Disease. Ceca from wildbirds were obtained during the spring 2004 turkey gun-huntseason. Intestinal samples from Missouri (five birds) andWisconsin (four birds) were isolated and frozen atDepartment of Natural Resources (DNR) registrationstations. Ceca from Iowa (four birds) were isolated by thehunters, frozen, and delivered to the laboratory at theNational Animal Disease Center. Wild birds were all free-range and presumably had no contact with antibiotics,coccidiostats or vaccines, although some contact withagricultural residues was possible (Table 1).

DNA Extraction Cecal samples (0.2 g) including cecal tissue,mucus, and lumen contents were suspended with lysingmatrix A (QBiogene, Carlsbad, CA) in 1-mL buffer L6(5.25 M guanidine hydrochloride, 50 mM Tris–HCl pH 6.4,20 mMEDTA, 1.3% (w/v) Triton X-100) [10]. Samples werelysed with a Fast Prep FP120 (Bio101, Savant) for 30 s atspeed 5 m/s DNAwas purified using the Qbiogene FastDNAKit as per the manufacturer’s instructions (Qbiogene,Carlsbad, CA). DNA was size fractionated in 1% agarosegels and fragments larger than 3 kb were purified using theQIAquick Gel Extraction Kit (Qiagen, Valencia, CA).

Polymerase Chain Reaction (Pcr) Amplification and 16SRibosomal Gene Library Construction Two 1495-clone16S ribosomal RNA (rRNA) gene libraries, one represent-ing 115 clones from each of 13 commercial turkeys and theother an equivalent number from 13 wild turkeys, wereprepared by cloning 16S genes into pNEB206a (NewEngland Biolabs, Beverly, MA). Bacterial 16S rRNA geneswere PCR-amplified for each bird as described previously,using modified primers 27F (GGGAAAGUAGRRTTTGA-TYHTGGYTCAG) and 1492R (GGAGACAUGB-TACCTTGTTACGACTT) [8, 29]. Thermocyclingreactions contained 50 mM Tris pH 8.3, 500 μg bovine

Table 1 Domestic and wild turkey samples 407 analyzed in this study

Bird ID Source Bird agea

MO1 Randolph Co, MO 2 yrMO2 Randolph Co, MO 3 yrMO3 Randolph Co, MO 3 yrMO4 Randolph Co, MO 4–5 yrMO5 Randolph Co, MO 1 yrIA1 Hamilton Co, IA 2 yrIA2 Hamilton Co, IA 2 yrIA3 Hardin Co, IA 3 yrIA4 Lucas Co, IA 1 yrWI1 Sauk Co, WI 2 yrWI2 Sauk Co, WI 3 yrWI3 Sauk Co, WI 2 yrWI4 Dane Co, WI 2 yrOA1 Farm A 18 wksOA2 Farm A 18 wksOA3 Farm A 18 wksOA4 Farm A 18 wksOA5 Farm A 18 wksSA1 Farm A 18 wksSA2 Farm A 18 wksSA3 Farm A 18 wksSA4 Farm A 18 wksSA5 Farm A 18 wksSB1 Farm B 18 wksSB2 Farm B 18 wksSB3 Farm B 18 wks

a Estimated from wild turkey spur length

Intestinal Microbiota of Turkeys 323323

serum albumin (BSA) per ml, 2.5 mM MgCl2, 250 μMeach diethylnitrophenyl thiophosphates (dNTPs), 400-nMprimers, and 1.75 U Taq per 20-μL reaction. Reactionparameters included a 5-min initial denaturation at 95°C.Cycling consisted of 1 min of 95°C denaturation, 30 s of48°C annealing and 40 s of 72°C elongation. Reactionswere finished with a 5-min elongation at 72°C. Genes wereamplified from fecal DNAs using the fewest number ofcycles possible to generate a visible product, generally 19–20 cycles. The 26 libraries included five from wild turkeysshot in Missouri (MO1–MO5), four from turkeys caught inIowa (IA1–IA4), four from turkeys caught in Wisconsin(WI1–WI4), five from domestic turkeys killed on-farm atfarm A (OA1–OA5), five domestic turkeys from farm Akilled at the abattoir (SA1–SA5) and three from farm Bturkeys killed at the abattoir (SB1–SB3).

Array Printing, Hybridization, and Statistical AnalysisOligonucleotide fingerprinting of rRNA genes (OFRG) wasperformed as described previously [46]. Briefly, the rRNAgenes from the 2,990 clones were PCR-amplified usingprimers UserOFRGFor2 and UserOFRGRev2 [8]. Reac-tions contained 50 mM Tris pH 8.3, 500 μg BSA per ml,2.5 mM MgCl2, 250 uM each dNTPs, 400 nM primers, and1.75 U Taq per 20-μL reaction. Thermocycling wasperformed for 35 cycles. Thermocycling parameters were5-min initial denaturation at 95°C followed by 35 cycles ofdenaturation at 95°C for 1 min and elongation at 72°C for2 min. Reactions were finished with a 5-min elongation at72°C. PCR products including those of 32 control cloneswere arrayed on nylon membranes as described previously[46]. Arrays were hybridized with a set of 37 10-nt probesthat were previously developed with simulated annealingand Lagrangian relaxation algorithms [11]. Two arrays werehybridized for each probe. Every array was then strippedand rehybridized with a universal probe [46]. Hybridizationintensities for duplicate membranes were averaged and thelog ratios of the discriminatory probes to the universalprobes calculated. Hybridization intensities were classifiedas 1 (positive hybridization event), 0 (negative hybridiza-tion event), or N (uncertain) using control clone hybridiza-tion intensities and Bayesian classification [25]. Uncertain(N) assignments were made when hybridization intensitieswere intermediate and could be classified as neither positivenor negative. Algorithms for fingerprint assignment using amodified Bayes classifier and dendrogram construction (theGreedy Clique Partition pAckage Tool, or GCPAT) can befound at: http://algorithms.cs.ucr.edu/OFRG/index.php.Clones determined to have more than 10 uncertain (N)classifications in their fingerprints were discarded fromfurther analysis as these fingerprints did not providetaxonomic information. Statistical comparisons of thelibraries were performed using GCPAT and SONS [40].

GCPAT calculated the richness (S), or number of uniqueOTU, the Shannon Index of diversity (H’) and Evenness(E), described as H’/ln S [22]. OFRG OTU numbers wereused as input for the SONS analysis program. Briefly, theoutput from the GCPAT program, which groups identicalfingerprints into OTU, was converted to the output formatfrom the DOTUR program [39]. As described previously,the SONS program was used to calculate SA,B ACE and SA,BChao1 estimates of shared richness, Jaccard and Sorensensimilarity indices, and the θYC and θN maximum likeli-hood estimations of community structure similarity. Princi-pal component analysis (PCA) was performed using SASv.8.2 (SAS Institute, Inc., Cary, NC).

Sequence Analysis Clones of interest (103) were sequencedusing primers M13F, M13R, 530F (GTGCCAGCMG-CCGCGG), and 907R (CCGTCAATTCMTTTRAGTTT)[29]. Sequences were aligned and edited using Vector NTI9.1.0 (Invitrogen, Carlsbad, CA). Sequences were com-pared to those in the public databases using National Centerfor Biotechnology Information (NCBI) Basic Local Align-ment Search Tool (BLAST) and the Ribosomal DatabaseProject II, and chimeric sequences were identified usingPintail http://www.cf.ac.uk/biosi/research/biosoft/Pintail/index.html) [4, 13]. Accession numbers for the abovesequences are EU009758–EU009861.

Dot Blot Quantification Dot blots were performed asdescribed previously [15]. Total DNA from each samplewas diluted in 0.1 N NaOH, 1 mM EDTA, and 250 ngapplied to positively charged nylon membrane (Amersham,Piscataway, NJ) using a Bio-Dot dot blotter (Bio-Rad,Hercules, CA). Bacterial DNA concentrations were equal-ized by hybridization with probe Eub338 (Table 2).Twenty-five nanograms of PCR-amplified Clostridiumgroup-IV, group-IX, and group-XIV 16S rRNA sequences,as well as those for Bacteroides, Faecalibacterium andMucispirillum, were applied to the membranes as controls[14]. 33P-labeled probes Eub338, Erec482, Clept1240,Bacto1080, Prop853, Faec645, and Deferr520 were dilutedin 5× Denhardt’s buffer: 5× standard saline citrate (SSC),0.5% sodium dodecyl sulfate (SDS), and 100 μg/ml salmonsperm DNA [1, 2, 18, 42]. Membranes were hybridizedovernight at 45°C and washed 2×15 min at 50°C withprimary wash buffer (0.5% SDS and 4× SSC) (Table 2) [3,18, 42]. For ease of use with a 50°C wash temperature,secondary wash buffers were optimized as describedpreviously (Table 2) [35]. Membranes were washed withappropriate secondary wash buffer for 30 min at 50°Cfollowed by exposure to a storage phosphor screen for 5 h.The image was visualized on a Typhoon 9410 imager(Amersham, Piscataway, NJ). Hybridization is described asthe ratio of signal from the specific probe divided by the

324 A. J Scupham et al.

signal from the universal bacterial probe Eub338. Reportedresults are averages of samples from each bird hybridizedseparately. Hybridization differences between the domesticand wild turkey total cecal DNAs for each probe weremeasured using the Student’s t test.

Results

Comparison of the Ribosomal Gene Libraries In all, 2,990clones from 26 animals were studied using OFRG. Of these2,990 clones, 544 contained fingerprints with ≥10 uncertain(N) assignments and were discarded from analysis. Clustercomparison of the domestic turkey and wild turkey clonelibraries indicated 685 operational taxonomic units (OTUs,here defined as identical fingerprints) in the domesticlibrary and 627 OTUs in the wild turkey library (Table 3,Supplemental Information Fig. 1). Of the 166 OTUscontaining ≥3 clones, 37% (61 OTU) were composedentirely of clones from the domestic library, whereas 26%(43 OTU) were uniquely wild turkey clones. The remaining37%, mostly OTUs containing >6 clones, were composedof both wild and domestic clones. Shannon’s diversityindex (H’) equaled 6.0 and evenness (E) equaled 0.9 forboth libraries independently. The abundance-based Jaccardand Sorensen’s similarity indices, for which a score of 1indicates identical libraries, were 0.09 and 0.17, for thedomestic and wild libraries, respectively. The two librariescombined had 1,294 OTUs, H ′=6.7 and E=0.9. Analyzingfor unique fingerprints with hybridization scores of 1, 0,and N equally weighted, 58% (standard error 13%) of thefingerprints from the domestic library were also found inthe wild library, whereas 52% (S.E. 10%) of the wildfingerprints were represented in the domestic library. In all,30% (S.E. 9%) of the 2,446 clones were shared. The S1,2ACE estimate of shared richness was 1,418 fingerprint typesand the S1,2 Chao estimate 750 types. Community structuresimilarity estimators θYC and θN yielded values of 0.29and 0.17, respectively [50, 51].

Principal component analysis was performed for the 26libraries of roughly 115 clones derived from individualbirds (Fig 1). The primary component identified a differ-ence between the clone libraries of the domestic and wildturkeys. The secondary component suggests a differencebetween the clone libraries of the turkeys from farms A(libraries OA1–5 and SA1–5) and B (libraries SB1–3).Differences were not noted for birds from farm A that werekilled on farm (OA1–5) vs those transported and killed atthe slaughterhouse (SA1–5), nor were differences apparentbetween the libraries of the wild birds (MO1–5, IA1–4,WI1–4).

Taxonomic Composition of Turkey Cecal Libraries Distri-bution of the clones into various taxa is described in Table 3.Clones representing the phylum Bacteroidetes includedgenera Alistipes, Prevotella, Bacteroides, and Megamonas.However, most clones (85%) were classifiable only to thephylum Bacteroidetes, class Bacteroidetes, or order Bacter-oidales. As described by the Ribosomal Database Project II(RDPII), 6% of the domestic Bacteroidetes clones clusteredwith sequenced clones having high (>99%) confidence asbelonging to the genus Alistipes, 2.6% belonging to thePrevotella, 4.5% belonging to the Bacteroides, and 4.4%belonging to the Megamonas [13]. Distribution of theAlistipes clones was very uneven, with 84% found in thedomestic turkey library. Conversely, 80% of the Megamonasclones derived from the wild turkey library.

The Firmicutes were the second most abundant taxo-nomic group, with clones grouping into the classes Bacilliand Clostridia. Forty-five percent of the Firmicutes clonesclustered with sequenced clones that could be classifiedonly as belonging to the order Clostridiales, and of these,72% were from the domestic turkey library. Within thisorder, the genera Anaerovorax, Dorea, Faecalibacterium,Subdoligranulum, Megasphaera, Phascolarctobacterium,and Papillibacter were identified. At 18% of the Firmicutesclones, Faecalibacterium was the predominant genus.Genera Lactobacillus, Enterococcus, and Streptococcus

Table 2 Dot blot probes to major bacterial taxa present in the turkey ceca

Target Sequence Wash buffer Reference

Univ1390 (Universal) GACGGGCGGTGTGTACAA 0.33X SSC [34]Eub338 (Bacterial universal) GCTGCCTCCCGTAGGAGT 0.75X SSC [2]Clept1240 (group IV) GTTTTRTCAACGGCAGTC 0.83X SSC [42]Faec645 (Poultry Faecalibacterium) CCTCTGCACTACTCAAGATACACa 0.33X SSC [45] ModifiedProp853 (group IX) ATTGCGTTAACTCCGGCAC 0.2X SSC [47]Erec482 (group XIV) GCTTCTTAGTCARGTACCG 0.63X SSC [18]Bacto1080 (Bacteroides) GCACTTAAGCCGACACCT 0.25X SSC [15]Deferr520 (Mucispirillum) GTGTTGTAAGTCATTAGT 0.33X SSC This work

a Italicized nucleotides indicate sequence divergence from the original publication

Intestinal Microbiota of Turkeys 325325

were identified as rare representatives of the Bacilli in theturkey cecum.

Classes β, δ, and γ-proteobacteria represented only asmall fraction of the clones, with genus Sutterella occurringmost frequently (2.6% of the combined domestic and wildturkey libraries). Within the phylum Deferribacteres, onlyclones distantly related (<96% sequence similarity) to

Mucispirillum schaedleri were identified. Finally, theAcinetobacter were represented by only two clones looselyrelated to the genus Eggerthella.

Clone sequences identified as bacteria, but otherwiseunclassifiable, represented 6.2% of the domestic turkeylibrary and 8.8% of the wild library. These clones clusteredwith other clones having low sequence similarity to the

Table 3 Taxonomic distribution of rRNA gene clones obtained by OFRG analysis of domestic and wild turkey ceca

Taxon No. of clones Nearest relative (GenBank accession no.) Percent similarity

Domestic Wild Total

Bacteroidetes 668 (52%)a 615 (56%) 1,283 (54%)Phylum Bacteroidetes 46 (4%) 21 (2%) 67Class Bacteroidetes 173 (13%) 114 (10%) 287Order Bacteroidales 372 (29%) 432 (39%) 804Genus Alistipes 42 (3%) 8 50 Alistipes putredinis (L16497) 90%Genus Prevotella 22 (2%) 13 (1%) 35 Prevotella loeschei (AY836508) 89%Genus Bacteroides 23 (2%) 38 (3%) 61 Bacteroides uniformis (AB247142) 93%Firmicutes 427 (33%) 351 (32%) 719 (30%)Phylum Firmicutes 7 6 13Class Bacilli 1 1Genus Lactobacillus 1 4 5 Lactobacillus aviarius (AB175732) 99%Genus Enterococcus 6 9 15 Enterococcus cecorum (AJ301827) 99%Genus Streptococcus 3 17 (2%) 20 Streptococcus alactolyticus (AF201899) 99%Class ClostridiaOrder Clostridiales 184 (14%) 80 (7%) 264Genus Anaerovorax 8 8 A. odorimutans (AJ251215) 89%Genus Dorea (XIVa) 5 12 (1%) 17 Clostridium sp. 14505 (AJ318890) 93%Genus Faecalibacterium (IV) 56 (4%) 73 (7%) 129 Faecalibacterium (AJ270469) 95%Genus Subdoligranulum (IV) 18 (1%) 3 21 S. variabile (AJ518869) 97%Family Acidaminococcaceae 2 15 (1%) 17Genus Megasphaera (IX) 0 32 (3%) 32 M. elsdenii (AY038994) 95%Genus Phascolarctobacterium 52 (4%) 21 (2%) 73 P. faecium (X72867) 94%Genus Papillibacter 72 (6%) 32 (3%) 104 Clostridium orbiscindens (AY730664) 96%Genus Megamonas (IX) 12 (1%) 47 (4%) 59 M. hypermegale (AJ420107) 99%Proteobacteria 63 (5%) 20 (2%) 83 (3%)Class BetaproteobacteriaGenus Sutterella 51 (4%) 10 (1%) 61 Sutterella stercoricanis (AJ566849) 98%Class DeltaproteobacteriaOrder Desulfovibrionales 5 8 13Class GammaproteobacteriaOrder PseudomonadalesGenus Acinetobacter 1 1 Acinetobacter parvus (AJ293691) 99%Order Enterobacteriales 1 1Genus Escherichia 6 1 7 Escherichia coli 100%Deferribacteres 51 (4%) 12 (1%) 63 (3%)Family Deferribacteraceae 51 12 63 Mucispirillum schaedleri (AY387670) 96%Actinobacteria 1 1 2Family Coriobacteriaceae 1 1 2 Eggerthella sinensis (AY321958) 89%Unclassified Bacteria 81 (6%) 96 (9%) 177 (7%)TOTAL 1,301 1,106 2,407# OTUs 685 627 1,294Shannon index (H’) 6.00 5.99 6.68Evenness (E) 0.92 0.93 0.93

a Percents indicate the ratio of clones within each library.

326 A. J Scupham et al.

phylum Bacteroidetes (83–86% confidence). BLAST anal-ysis of these clones indicated low (≤85%) sequencesimilarity to uncultured intestinal clones from varioussources. These clones indicate dominant, poorly definedgroups of organisms.

As indicated by the principal component analysis,taxonomic distribution of OFRG clones based on individualbirds indicated bird-to-bird variability, even within cohab-iting animals (Fig. 2). However, the overall trends observedwith the pooled data reported in Table 3 were maintained,with greater proportions of Deferribacteres and Proteobac-teria apparent in the domestic animals.

Quantification of Cecal Tax Probes for general intestinalmicrobe groups were used in dot blot hybridization assays:Bacteroidetes, Clostridia groups IV, IX, and XIV (Table 4)[14]. In addition, probes for the Faecalibacteria andDeferribacteres were developed. The domestic turkey cecalDNAs contained more clostridia group XIV target sequen-ces (39.7%) than did the wild turkey cecal DNAs (27.3%),

p=0.013 (Table 4). Significant differences were not seenbetween the domestic and wild turkey samples for theremaining taxa, but in general Deferribacteres and Clos-tridiales proportions were higher, and Bacteroidetes lower,than the proportions identified in the clone libraries. ProbeClept1240 (Collins group IV) was determined to cross-reactwith a subset of the Clostridium group XIV. In addition,probe Faec645 was designed to detect poultry-specificFaecalibacterium sequences and thus reacts with a subsetof Clostridium group IV sequences.

Discussion

Oligonucleotide fingerprinting of rRNA genes (OFRG) wasused to identify 16S rRNA clones from 26 birds for thepurpose of describing and identifying differences betweenthe cecal microbiota of commercially raised and wildturkeys. This survey indicated that the bacterial communi-ties were dominated by Bacteroidetes and Firmicutes.Additional taxa included Actinobacteria, Deferribacteres,and β, δ and γ-proteobacteria. All the major taxa werepresent in both the domestic and wild turkey libraries,although not always in equal ratios. Seventy-three percentof the clones could not be taxonomically assigned to anyknown genus; however, a few groups did have significantsimilarity to cultured intestinal species. A large group ofClostridiaceae clones clustered with butyrate-producingFaecalibacterium, a genus recently identified as taxonom-ically closer to the Clostridium leptum group (IV) than tothe Fusobacteria [16]. The Faecalibacteria are common inhuman feces and a group-specific probe was previouslydesigned for their enumeration [45]. Using clone sequencesfrom the current work and a previous publication, a poultry-specific Faecalibacterium probe was developed from theexisting human probe sequence [41]. BLAST analysisindicated that the new probe was 100% identical only to

0%

20%

40%

60%

80%

100%

OA

1

OA

2

OA

3

OA

4

OA

5

SA

1

SA

2

SA

3

SA

4

SA

5

SB

1

SB

2

SB

3

MO

1

MO

2

MO

3

MO

4

MO

5

IA1

IA2

IA3

IA4

WI1

WI2

WI3

WI4

Bacteria

Deferribacteres

Proteobacteria

Firmicutes

Bacteroidetes

Figure 2 Relative abundances of bacterial taxonomic groups repre-sented by clones from individual turkeys. Principal components 1 and2 combined account for 18% of the variance. OA, birds killed on farm

at farm A; SA, birds from farm A killed at the abattoir; SB, birds fromfarm B killed at the abattoir; MO, wild birds shot in Missouri; IA wildbirds shot in Iowa; WI, wild birds shot in Wisconsin

-12

-10

-8

-6

-4

-2

0

2

4

6

8

-6 -4 -2 0 2 4 6 8

Prin1 (11%)

Pri

n2

(7%

)

OA

SA

SB

MO

IA

WI

Figure 1 Principal component analysis of the clone librariesgenerated for each of the 26 turkeys. OA, birds killed on farm atfarm A; SA, birds from farm A killed at the abattoir; SB, birds fromfarm B killed at the abattoir; MO, wild birds shot in Missouri; IA wildbirds shot in Iowa; WI, wild birds shot in Wisconsin

Intestinal Microbiota of Turkeys 327327

archived Faecalibacterium sequences obtained from poul-try sources. Future analysis of Faecalibacteria isolates fromvarious hosts will be necessary to determine whetherFaec645 is truly poultry specific, and thus useful as asource-specific identifier of fecal contamination of water-ways. Megamonas hypermegale clones represent a secondlarge group found predominantly in the wild turkey library.It has been found in the mucosa of the human gut and inpoultry, and production of propionic and acetic acids by thisspecies has been thought to inhibit Salmonella colonizationof the poultry intestine [6, 37]. Historically considered amember of the Bacteroides, this group has also recentlybeen suggested as a member of the Clostridiales [32]. In ourOFRG analyses, the Megamonas clones cluster tightly withMegasphaera elsdenii clones and clones from the familyAcidaminococcaceae; clustering of these 16S rRNAsequences has also been noted previously [19]. Neverthe-less, the Megamonas clones described in this work havebeen classified with the Bacteroidetes in Table 2 to conformto the current RDP II taxonomic structure. Megasphaeraelsdenii is commonly found in the gastrointestinal tracts ofruminants, pigs, and humans. M. elsdenii is well-known asa lactate reducer in the bovine rumen, important forprevention of acidosis in grain-fed animals. M. elsdeniipopulations are affected by diet, as they were suppressed inpigs fed low protein and promoted in cattle fed grain ratherthan grass [7, 28]. In the present studies, Megasphaeraclones were found solely in the wild turkey library.Whether the diets of the wild birds, generally composedof seeds, insects, worms, berries, and grasses, is higher inprotein than the corn/soy diets of the domestic birds isunknown.

Differences between the domestic and wild turkeylibraries were readily observed in relation to relativelyminor components of the communities. Nine sequencedclones represented clusters containing 62 domestic and 10wild clones, and had between 95% and 90% similarity tothe order Burkholderiales and Sutterella species [27].Sutterella species have been identified in human and caninefeces, can be co-cultured from these samples with the food-

borne pathogen Campylobacter, and may be involved ingastrointestinal disorders [17]. It is possible that Sutterellais a direct competitor for Campylobacter in the intestinalecosystem. The OFRG cluster associated with 16S rRNAgenes from the genus Deferribacteriaceae included cloneswith sequence similarity to Mucispirillum schaedleri andthe genus Geovibrio. These spiral organisms have beenfound to reside in the mucus layer of the mammalianintestinal tract, the human oral cavity, and in aquaticecosystems [24, 31, 36]. In 1979, Hanssen described thepresence of spirochaetes in the intestines of wild grouse,which may have been the first description of these microbesin the gallinaceous gut [20]. Geovibrio species detected inhog feces degrade volatile fatty acids such as butyrate viaFe(III) reduction [12]. The health benefits of butyrate in theintestine have been widely studied, thus this understudiedgroup may prove important for intestinal health [48].

Dot blot quantification of the major taxa in the turkeyintestine was performed using four previously publishedand two new probes. A probe specific for the turkeyDeferribacteres was developed and BLAST analysis andcomparison of the Deferr520 probe to cloned sequencesfailed to identify undesirable cross-reactivity with turkeyintestinal samples. Relative to clone numbers classified byOFRG, dot blot hybridization using probe Deferr520indicated a much larger population of Deferribacteralesthan anticipated. The reason for this discrepancy isunknown and further analyses will need to be performedto determine whether cross-reactivity or PCR-derivedunderrepresentation in the clone libraries caused thisobservation. A probe specific for Faecalibacterium wasadapted from the literature for the turkey Faecalibacteria.BLAST analysis of the modified Faec645 probe indicated100% matches only to poultry (chicken and turkey)Faecalibacterium intestinal clones. Analysis of turkeyClostridium group IV clones from a previous publicationindicated 118 clones (70%) contained the binding sequen-ces for both Clept1240 and Faec645, 33 (20%) having thebinding site for Faec645 only and 17 (10%) with only theClept1240 binding site [41]. Likewise, an overlap was seen

Table 4 Average composition of bacterial 16S rDNA sequences as determined by dot blot hybridization

Probe Composition of bacterial 16S sequences Percent (St Dev)

Domestic Wild

Deferr520 (Mucispirillum) 10.8 (5.4) 10.7 (4.1)Faec645 (IV) 16.1 (12.0) 19.9 (9.9)Clept1240 (IV) 6.6 (2.8) 7.2 (3.1)Prop853 (IX) 12.3 (4.6) 14.9 (4.1)Erec482 (XIV)a 39.7 (11.8) 27.3 (11.6)Bacto1080 15.8 (6.1) 16.4 (3.8)

a p=0.013

328 A. J Scupham et al.

between clones containing target sequences for probesErec482 (group XIV) and Clept1240, but at very lowfrequency. Analysis of Bacteroidetes sequences from aprevious turkey intestinal microbiota study indicated 22%of Bacteroidetes clones in a random library did not containbinding sites for probe Bacto1080 [41].

Statistical analysis of the OFRG library results mayunderestimate actual diversity. By definition, OFRG finger-prints are coded representations of complete 16S rRNAsequences, thus much information is lost. Addition of theuncertain, or N, values causes further loss of informationwhile simultaneously greatly enlarging the pool of potentialfingerprints. With these caveats in mind, the S1,2 ACE andS1,2 Chao richness estimates of 1,418 and 750 refer to thenumber of OTUs in each sample, not to the number ofspecies. Measurement of the number of OTUs for thecombined libraries (1,294 OTUs) as slightly less than thesum of the number of OTUs for the individual libraries(1312 OTUs), in conjunction with high Shannon’s diversity,high evenness, and low Jaccard and Sorensen’s similarityindices, indicated very little fingerprint-type overlap andlow community structure similarity. Comparison of clonelibraries via SONS, a program that calculates a variety ofparameters, indicated 30% of the fingerprints were sharedbetween the two libraries [40]. The parameters θYC andθN, estimating community structure similarity, indicatedlow probability for observing the same fingerprint whenone fingerprint is randomly chosen from each library (θYC)and low probability that a given fingerprint present in onelibrary is present in both (θN) [50, 51]. Principal compo-nent analysis supported the calculated differences betweenthe domestic and wild turkey libraries, but also indicated adifference between the sublibraries from birds fromdifferent farms. This minor difference likely derives fromFarm B-specific clusters within the Bacteroidales andDeferribacteres. As clones from Farms A and B werecomponents of a single library and analyzed simultaneous-ly, these differences cannot be ascribed to many of thestochastic factors described elsewhere in this discussion.Thus, these differences may be the result of differentialantibiotic regimens specific to the individual flocks.Alternatively, the differences could result from initialcolonizers directing the ultimate composition of theintestinal microbiota [21].

Host age, genotype, diet, exposure to antibiotics, andinitial exposure to bacterial consortia in the environment allmay have played a role in the differences observed betweenthe domestic and wild turkey libraries. As described inTable 1, the domestic birds were killed at 18 weeks of age,whereas the ages of the wild birds, estimated from spurlength, were between 1 and 5 years. Previous publicationsexamining cecal community changes in poultry identifiedshifts throughout the production cycle [23, 30, 41]. In

addition, intestinal community shifts have been noted forelderly humans, including decreased Bacteroides andBifidobacteria and increased Fusobacteria and Clostridia[49]. In addition to age, antibiotics likely played asubstantial role in the development of the intestinal micro-biota of the domestic turkeys. These animals were raisedaccording to current accepted commercial practices, includ-ing vaccinations against pneumovirus and Newcastle’sDisease, and lifelong treatment with the growth promotantvirginiamycin. In addition, animals were therapeuticallytreated with arsenicals, coccidiostats, anthelminthics, andpenicillin G. The exposure of the wild birds to equivalentcompounds is unknown, although some exposure in highlyagricultural areas of the Midwest cannot be excluded.

Possible laboratory artifacts resulting in library differ-ences should not be overlooked. The domestic and wildturkey libraries were made, printed, and hybridized sepa-rately from one another, providing an opportunity forvariability. In addition, a complete set of control cloneswas not available. Therefore, a set of random clones wassequenced and included on both arrays (identified asControl on the dendrogram, Supplemental Information,Fig 1). This set of controls proved insufficient, so moreclones from each array were sequenced and included ascontrols during data analysis. This discrepancy may haveallowed inconsistency in the assignment of 0, 1, and N tothe fingerprints. Random differences in 16S rRNA se-quence amplification could also contribute to discrepanciesbetween the two libraries; however, each library wascomposed of 13 sublibraries, each derived from one of the13 different birds. Stochastic preferential amplification ofrRNA sequences should thus be minimized. Even with theknown limitations of the method, the results reported hereidentify Sutterella and Megasphaera species as candidatesfor competitive exclusion studies. In addition, the resultsprovide valuable insight into a poorly described ecosystemand provide the foundation for further studies into host–microbe and microbe–microbe interactions that affectanimal health, nutrition, and food safety.

Acknowledgments The authors gratefully acknowledge the techni-cal assistance of Jennifer A. Jones for all stages of data generation. Wealso thank David P. Alt and Karen Hollum for sequencing services.

Disclaimer Mention of trade names or commercial products in thisarticle is solely for the purpose of providing specific information anddoes not imply recommendation or endorsement by the U.S.Department of Agriculture.

References

1. Amann R, Ludwig W (2000) Ribosomal RNA-targeted nucleicacid probes for studies in microbial ecology. FEMS MicrobiolRev 24:555–565

Intestinal Microbiota of Turkeys 329329

2. Amann RI, Binder BJ, Olson RJ, Chisholm SW, Devereux R,Stahl DA (1990) Combination of 16S rRNA-targeted oligonucle-otide probes with flow cytometry for analyzing mixed microbialpopulations. Appl Environ Microbiol 56:1919–1925

3. Amann RI, Krumholz L, Stahl DA (1990) Fluorescent-oligonu-cleotide probing of whole cells for determinative, phylogenetic,and environmental studies in microbiology. J Bacteriol 172:762–770

4. Ashelford KE, Chuzhanova NA, Fry JC, Jones AJ, Weightman AJ(2005) At least 1 in 20 16S rRNA sequence records currently heldin public repositories is estimated to contain substantial anoma-lies. Appl Environ Microbiol 71:7724–7736

5. Barnes EM (1979) The intestinal microflora of poultry and gamebirds during life and after storage. Address of the president of theSociety for Applied Bacteriology delivered at a meeting of thesociety on 10 January 1979. J Appl Bacteriol 46:407–419

6. Barnes EM, Impey CS, Stevens BJ (1979) Factors affecting theincidence and anti-salmonella activity of the anaerobic caecal floraof the young chick. J Hyg (Lond) 82:263–283

7. Benno Y, Endo K, Suzuki K, Mitsuoka T, Namioka S (1985) Useof nonprotein nitrogen in pigs: effects of dietary urea on theintestinal microflora. Am J Vet Res 46:959–962

8. Bent E, Yin B, Figueroa A, Ye X, Fu Q, Liu Z, Chrobak M, JeskeD, Jiang T, Borneman J (2006) Development of a 9,6000 clonearray for oligonucleotide fingerprinting of rRNA genes: utilizationto compare four different soil DNA extraction methods. JMicrobiol Methods 67:171–180

9. Bielke LR, Elwood AL, Donoghue DJ, Donoghue AM, NewberryLA, Neighbor NK, Hargis BM (2003) Approach for selection ofindividual enteric bacteria for competitive exclusion in turkeypoults. Poult Sci 82:1378–1382

10. Boom R, Sol CJ, Salimans MM, Jansen CL, Wertheim-van DillenPM, van der Noordaa J (1990) Rapid and simple method forpurification of nucleic acids. J Clin Microbiol 28:495–503

11. Borneman J, Chrobak M, Della Vedova G, Figueroa A, Jiang T(2001) Probe selection algorithms with applications in the analysisof microbial communities. Bioinformatics 17 Suppl 1:S39–S48

12. Coates JD, Cole KA, Michaelidou U, Patrick J, McInerney MJ,Achenbach LA (2005) Biological control of hog waste odorthrough stimulated microbial Fe(III) reduction. Appl EnvironMicrobiol 71:4728–4735

13. Cole JR, Chai B, Farris RJ, Wang Q, Kulam SA, McGarrell DM,Garrity GM, Tiedje JM (2005) The Ribosomal Database Project(RDP-II): sequences and tools for high-throughput rRNA analysis.Nucleic Acids Res 33:D294–D296

14. Collins MD, Lawson PA, Willems A, Cordoba JJ, Fernandez-Garayzabal J, Garcia P, Cai J, Hippe H, Farrow JA (1994) Thephylogeny of the genusClostridium: proposal of five new genera andeleven new species combinations. Int J Syst Bacteriol 44:812–826

15. Dore J, Sghir A, Hannequart-Gramet G, Corthier G, Pochart P(1998) Design and evaluation of a 16S rRNA-targeted oligonu-cleotide probe for specific detection and quantitation of humanfaecal Bacteroides populations. Syst Appl Microbiol 21:65–71

16. Duncan SH, Hold GL, Harmsen HJ, Stewart CS, Flint HJ (2002)Growth requirements and fermentation products of Fusobacteriumprausnitzii, and a proposal to reclassify it as Faecalibacteriumprausnitzii gen. nov., comb. nov. Int J Syst Evol Microbiol52:2141–2146

17. Engberg J, On SL, Harrington CS, Gerner-Smidt P (2000)Prevalence of Campylobacter, Arcobacter, Helicobacter, andSutterella spp. in human fecal samples as estimated by areevaluation of isolation methods for Campylobacters. J ClinMicrobiol 38:286–291

18. Franks AH, Harmsen HJ, Raangs GC, Jansen GJ, Schut F,Welling GW (1998) Variations of bacterial populations in humanfeces measured by fluorescent in situ hybridization with group-

specific 16S rRNA-targeted oligonucleotide probes. Appl EnvironMicrobiol 64:3336–3345

19. Gophna U, Sommerfeld K, Gophna S, Doolittle WF, Veldhuyzenvan Zanten SJ (2006) Differences between tissue-associatedintestinal microfloras of patients with Crohn’s disease andulcerative colitis. J Clin Microbiol 44:4136–4141

20. Hanssen I (1979) A comparison of the microbiological conditionsin the small intestine and caeca of wild and captive willow grouse(Lagopus lagopus lagopus). Acta Vet Scand 20:365–371

21. Hooper LV, Wong MH, Thelin A, Hansson L, Falk PG, Gordon JI(2001) Molecular analysis of commensal host-microbial relation-ships in the intestine. Science 291:881–884

22. Hughes JB, Bohannan BJM (2004) Application of ecologicaldiversity statistics in microbial ecology. In: Kowalchuk GA, deBruijn FJ, Head IM, Akkermans ADL, van Elsasz JD (eds)Molecular microbial ecology manual, vol. 2, 2nd ed. KluwerAcademic Press, The Netherlands, pp 1321–1344

23. Hume ME, Kubena LF, Edrington TS, Donskey CJ, Moore RW,Ricke SC, Nisbet DJ (2003) Poultry digestive microflorabiodiversity as indicated by denaturing gradient gel electrophore-sis. Poult Sci 82:1100–1107

24. Hutter G, Schlagenhauf U, Valenza G, Horn M, Burgemeister S,Claus H, Vogel U (2003) Molecular analysis of bacteria inperiodontitis: evaluation of clone libraries, novel phylotypes andputative pathogens. Microbiology 149:67–75

25. Jampachaisri K, Valinsky L, Borneman J, Press SJ (2005)Classification of oligonucleotide fingerprints: application formicrobial community and gene expression analyses. Bioinfor-matics 21:3122–3130

26. Jernberg C, Lofmark S, Edlund C, Jansson JK (2007) Long-termecological impacts of antibiotic administration on the humanintestinal microbiota. The International Society for MicrobialEcology Journal 1:56–66

27. Jousimies-Somer H (1997) Recently described clinically impor-tant anaerobic bacteria: taxonomic aspects and update. Clin InfectDis 25 Suppl 2:S78–S87

28. Krause DO, Smith WJ, Conlan LL, Gough JM, Williamson MA,McSweeney CS (2003) Diet influences the ecology of lactic acidbacteria and Escherichia coli along the digestive tract of cattle:neural networks and 16S rDNA. Microbiol 149:57–65

29. Lane DJ (1991) 16S/23S rRNA Sequencing. In: Stackebrandt E,Goodfellow M (eds) Nucleic acid techniques in bacterialsystematics. Wiley, New York, pp 115–175

30. Lu J, Idris U, Harmon B, Hofacre C, Maurer JJ, Lee MD (2003)Diversity and succession of the intestinal bacterial community of thematuring broiler chicken. Appl Environ Microbiol 69:6816–6824

31. Miroshnichenko ML, Bonch-Osmolovskaya EA (2006) Recentdevelopments in the thermophilic microbiology of deep-seahydrothermal vents. Extremophiles 10:85–96

32. Morotomi M, Nagai F, Sakon H (2007) Taxonomic note: genusMegamonas should be placed in the lineage of Firmicutes;Clostridia; Clostridiales; Acidaminococcaceae; Megamonas. Int JSyst Evol Microbiol 57:1673–1674

33. Nurmi E, Rantala M (1973) New aspects of Salmonella infectionin broiler production. Nature 241:210–211

34. Pace NR, Stahl D, Lane DJ, Olsen GJ (1986) The analysis ofnatural microbial populations by ribosomal RNA sequences. AdvMicrob Ecol 9:51–55

35. Raskin L, Stromley JM, Rittmann BE, Stahl DA (1994) Group-specific 16S rRNA hybridization probes to describe natural commu-nities of methanogens. Appl Environ Microbiol 60:1232–1240

36. Robertson BR, O’Rourke JL, Neilan BA, Vandamme P, On SL,Fox JG, Lee A (2005) Mucispirillum schaedleri gen. nov., sp.nov., a spiral-shaped bacterium colonizing the mucus layer of thegastrointestinal tract of laboratory rodents. Int J Syst EvolMicrobiol 55:1199–1204

330 A. J Scupham et al.

37. Salanitro JP, Blake IG, Muirhead PA (1974) Studies on the cecalmicroflora of commercial broiler chickens. Appl Microbiol28:439–447

38. Schales K, Gerlach H, Kosters J (1993) Investigations on theaerobic flora and Clostridium perfringens in fecal specimens fromfree-living and captive capercaillies (Tetrao urogallus L., 1758).Zentralbl Veterinarmed B 40:469–477

39. Schloss PD, Handelsman J (2005) Introducing DOTUR, a computerprogram for defining operational taxonomic units and estimatingspecies richness. Appl Environ Microbiol 71:1501–1506

40. Schloss PD, Handelsman J (2006) Introducing SONS, a tool foroperational taxonomic unit-based comparisons of microbialcommunity memberships and structures. Appl Environ Microbiol72:6773–6779

41. Scupham AJ (2007) Succession in the intestinal microbiota ofpreadolescent turkeys. FEMS Microbiol Ecol 60:136–147

42. Sghir A, Gramet G, Suau A, Rochet V, Pochart P, Dore J (2000)Quantification of bacterial groups within human fecal flora byoligonucleotide probe hybridization. Appl Environ Microbiol66:2263–2266

43. Souza MR, Moreira JL, Barbosa FH, Cerqueira MM, Nunes AC,Nicoli JR (2007) Influence of intensive and extensive breeding onlactic acid bacteria isolated from Gallus gallus domesticus ceca.Vet Microbiol 120:142–150

44. Stern NJ, Svetoch EA, Eruslanov BV, Perelygin VV, MitsevichEV, Mitsevich IP, Pokhilenko VD, Levchuk VP, Svetoch OE, SealBS (2006) Isolation of a Lactobacillus salivarius strain and

purification of its bacteriocin, which is inhibitory to Campylo-bacter jejuni in the chicken gastrointestinal system. AntimicrobAgents Chemother 50:3111–3116

45. Suau A, Rochet V, Sghir A, Gramet G, Brewaeys S, Sutren M,Rigottier-Gois L, Dore J (2001) Fusobacterium prausnitzii andrelated species represent a dominant group within the human fecalflora. Syst Appl Microbiol 24:139–145

46. Valinsky L, Scupham AJ, Della Vedova J, Liu Z, Figuerosa A,Jampachaisri K, Yin B, Press J, Jiang T, Borneman J (2004)Oligonucleotide Fingerprinting of Ribosomal RNA Genes (OFRG).In Kowalchuk GA, de Bruijn FJ, Head IM, Akkermans ADL, vanElsas JD (eds) Molecular microbial ecology methods, vol. 1, 2nd ed.Kluwer Academic Press, The Netherlands, pp 569–585

47. Walker AW, Duncan SH, McWilliam Leitch EC, Child MW, FlintHJ (2005) pH and peptide supply can radically alter bacterialpopulations and short-chain fatty acid ratios within microbialcommunities from the human colon. Appl Environ Microbiol71:3692–3700

48. Wong JM, de Souza R, Kendall CW, Emam A, Jenkins DJ (2006)Colonic health: fermentation and short chain fatty acids. J ClinGastroenterol 40:235–243

49. Woodmansey EJ (2007) Intestinal bacteria and ageing. J ApplMicrobiol 102:1178–1186

50. Yue JC, Clayton MK (2005) A similarity measure based onspecies proportions. Commun Stat 34:2123–2131

51. Yue JC, Clayton MK, Lin FC (2001) A nonparametric estimatorof species overlap. Biometrics 57:743–749

Intestinal Microbiota of Turkeys 331331