OUP user menu

Succession in the intestinal microbiota of preadolescent turkeys

Alexandra J. Scupham
DOI: http://dx.doi.org/10.1111/j.1574-6941.2006.00245.x 136-147 First published online: 1 April 2007


In the present study, automated ribosomal intergenic spacer analysis (ARISA), library sequence analysis, real-time PCR detection of Bacteroides uniformis and Campylobacter coli and dot-blot hybridizations of Clostridiaceae were used to identify trends in microbial colonization of the ceca of male turkeys. Two separate trials were performed with six and five birds, respectively. ARISA community profiles identified a period of community transition at week 12 of age in both trials. A significant increase of Ca. coli was also detected at week 12 in one trial, suggesting a possible correlation between microbiota destabilization and pathogen prevalence. Libraries of ribosomal small subunit 16S genes representing weeks 9, 11, 12 and 14 of both trials were sequenced. Whereas fingerprint and sequence analyses indicated significant differences in the species composition between the two trials, in general sequence library and dot-blot analyses indicated that Clostridia-like species decreased in prevalence over time. While B. uniformis prevalence in the two trials rose from 7% and 0% of the library clones at week 9 to 84% and 79% at week 11, real-time PCR did not support these results, with only approximately twofold and sixfold increases in internal transcribed spacer copy numbers observed.

  • ecology
  • poultry
  • Bacteroides uniformis
  • Campylobacter


With the development of molecular microbial ecology techniques, the complexity of intestinal microbial communities and the functions they perform are beginning to be understood. Correlations have been made between intestinal microbiota and diseases such as Crohn's disease and obesity (Backhed et al., 2004; Cobrin & Abreu, 2005; Ley et al., 2005). The worldwide reported allergy increase may also be the result of suppressed intestinal community development associated with high levels of antibiotic use (Kalliomaki & Isolauri, 2003). Intestinal microbial ecology also affects animal health and nutrition, and thus food safety.

Poultry ceca are primary colonization sites for human and animal pathogens such as Campylobacter, Salmonella, Listeria, Yersinia, Escherichia coli and Clostridium perfringens, and rupture of these organs during slaughter plays a role in the contamination of food products (Hargis et al., 1995). Culture-based studies of cecal bacterial communities suggest full microbial establishment may occur between 30 and 42 days, although these studies provide only a rough description of the microorganisms comprising the rich intestinal community (Barnes & Impey, 1972; Mead, 1989). In the human intestine, molecular methods have detected c. 800 species or >7000 unique sequences and an estimated 90% of avian gut microbial types are currently undescribed (Apajalahti et al., 2004; Backhed et al., 2005). Initial cecal communities derived from shell contaminants are known to consist predominantly of enterococci, coliforms and clostridia (Coates & Fuller, 1977). By day 4, lactobacilli become a significant component of the microbiota and the communities culminate in clostridia, sporomusa, enterics and bacteroides (Zhu et al., 2002). In addition to age, factors such as host diet, stress and antibiotic use cause microbial populations to change (Maciorowski et al., 1997; Craven, 2000; Wesley et al., 2005). It was found that in commercial broiler flocks subjected to antibiotic treatment, species richness stabilizes at 10 days of age, but community composition continues to change with a period of stabilization during rapid skeletal growth followed by further changes during growout (van der Wielen et al., 2002; Hume et al., 2003; Lu et al., 2003). Throughout community development, relative proportions of clostridia and bacteroides are stable, while lactobacilli and proteobacteria decrease in numbers after the first week and a significant increase in bifidobacteria is seen between days 4 and 14 (Lu et al., 2003; Amit-Romach et al., 2004).

The objective of the present study was to expand the current understanding of cecal community development in the turkey, an animal that is grown to 18 weeks as contrasted to the 6 weeks standard in the broiler industry. Although photoperiod strongly affects the onset of puberty, both male turkeys and chickens produce detectable testosterone by 19–22 weeks of age (Johnson, 1986; Vanmontfort et al., 1995; Bacon et al., 2000). Thus, previous studies of chicken microbial community development included only immature animals. The current study may be the first to track community development during the turkey growout phase, and to do so by examining individual birds rather than flocks. Automated ribosomal intergenic spacer analysis (ARISA) was used to describe microbial community changes over time and identify periods of interest. ARISA involves amplification of the intergenic spacers between the rrn genes (ITS). These sequences are highly divergent in length as well as sequence; use of fluorescently labeled primers to amplify the ITS from environmental samples, followed by amplicon separation on a sequence analyser, generates a pattern of peaks descriptive of the microbial community. ITS amplification from environments sampled over time or space thus allows identification of microbial community shifts. Sequence analysis of ribosomal small-subunit genes (SSU) were then used to identify the bacterial species. Real-time PCR analysis was used to quantify changes in Bacteroides uniformis and Campylobacter coli numbers and dot-blot analyses were used to show relative amounts of Clostridiaceae from transitional samples.

Materials and methods

Animals, feed and housing, and sample collection

Male Hybrid Converter turkeys (Midwest Turkey Hatchery) in two trials were banded for identification and housed together from day of hatch through to 18 weeks of age. The two time-course trials, Trial 1 and Trial 2, were performed spring/summer 2004 and summer/autumn 2005, respectively. Six birds comprised Trial 1 and five comprised Trial 2.

Birds were given water and feed ad libitum throughout the trials. From day of hatch through to 6 weeks of age the birds received Kent High Flyer 28 containing a 28% soybean and fish meal protein formulation, and thereafter Kent High Flyer 22 containing a 22% soybean and fish meal protein formulation (Kent Feeds, Inc.). Feed met or exceeded critical nutrient requirements recommended by the National Research Council and neither feed contained antibiotics, growth promotants or coccidiostats (National Research Council, 1994). Heat was supplied to the young birds with 150-W ceramic infrared heat emitters so animals could modulate personal temperatures between 85 and 95°C (Zoo Med Laboratories, Inc.). Incandescent lighting was supplied on a 12-h diurnal cycle.

Cecal feces were collected weekly by sampling the centers of droppings with sterile tongue depressors. Diurnal lighting allowed cecal feces collection in the 30 min after the barn was illuminated. Samples were immediately frozen on dry ice in sterile 1.5-mL tubes and stored at −80°C until DNA was extracted. DNAs were extracted in randomized groups, in periods immediately following the end of each timecourse.

Studies were conducted with approval of and in accordance with the instructions of the institutional animal care and use committee.

Total DNA isolation and ARISA

Total DNA was isolated from cecal samples with the Qbiogene FastDNA Kit using buffer CLS-TC and lysing matrix A (QBiogene). DNA was extracted as per the manufacturer's instructions, with bead beating in a FastPrep FP120 (QBiogene) for 30 s at 5 m s−1.

Bacterial ARISA was performed as described previously using the primer set ITSF and ITSReub (Table 1) (Borneman & Triplett, 1997; Fisher & Triplett, 1999; Cardinale et al., 2004). Primer ITSReub was 5′-end labeled with the FAM fluorochrome (Operon). Each 20-μL PCR reaction mixture contained 3.8 ng template DNA. Together with MapMarker 1000 size standard (BioVentures), PCR products were separated on an ABI3100 capillary sequencer for 167 min at 12.2 kV, using POP6 polymer and a 50-cm array (Applied Biosystems). Amplicon sizes were identified using the Local Southern size calling method with genemapper v3.5 software. The Dice coefficient paired with the neighbor-joining (NJ) algorithm of mega 3.0 was used to compare the ARISA profiles and prepare dendrograms (Fisher & Triplett, 1999; Simpson et al., 1999; McCracken et al., 2001; Hume et al., 2003; Kumar et al., 2004).

View this table:

Primers and probes used

Primer/probe namePrimer/probe sequence (5′–3′)Reference
ITSReubGCCAAGGCATCCACCCardinale et al. (2004)
16F2AGTCGTAACAAGGTARCCGTAKuwahara et al. (2001)
23R4GGGTTBCCCCATTCGGKuwahara et al. (2001)
Erec482GCTTCTTAGTCARGTACCGFranks et al. (1998)
Clept1240GTTTTRTCAACGGCAGTCSghir et al. (2000)
Euk1397TAGAAAGGGCAGGGAHicks et al. (1992)
Eub338GCTGCCTCCCGTAGGAGTAmann et al. (1990)

Sequence analysis

Eight bacterial 16S ribosomal SSU libraries representing weeks 9, 11, 12 and 14 of Trials 1 and 2 were constructed. These weeks were chosen to encompass the community transition described by the ARISA results. Amplification reactions (20 μL) contained the following reagents: 50 mM Tris (pH 8.3), 500 μg mL−1 bovine serum albumin (BSA), 2.5 mM MgCl2, 250 μM of each dNTP, 400 nM of each primer BACfOFRGpUSER and BACrOFRGpUSER (Table 1) (Operon), 3.8 ng template and 5 U Taq DNA polymerase (Roche Applied Science; Bent et al., 2006). The primers are modified versions of universal bacterial primers 27f and 1392r (Lane, 1991). Cycling parameters were: 94°C for 5 min; 35 cycles of 94°C for 30 s, 48°C for 40 s and 72°C for 1 min; final extension of 72°C for 2 min. Gel-purified products were cloned into pNEB206a (New England Biolabs). Clones were sequenced using primers M13F, M13R, 530F and 907R (Table 1) (Lane, 1991). Sequences were compared with the public databases using NCBI blast and the Ribosomal Database Project II (RDP II) (Cole et al., 2005). Chimeric sequences were identified using Pintail (http://www.cf.ac.uk/biosi/research/biosoft/Pintail/index.html) (Ashelford et al., 2005). Sequence clustering was performed using clustalx, and Jukes–Cantor distance matrices constructed from the phylip dnadist program were input to the libshuff program to detect significant differences between the libraries (Thompson et al., 1997; Singleton et al., 2004; Felsenstein, 2005). NJ, parsimony and maximum-likelihood phylogenetic analyses from the phylip package were used to cluster B. uniformis sequences. DOTUR assigned sequences to operational taxonomic units (OTUs) at 1% sequence dissimilarity and generated the Shannon–Weaver (H′) and Simpson (D) diversity indices for the eight libraries (Schloss & Handelsman, 2005). Evenness (E) was calculated as described previously (Hughes & Bohannan, 2004).

Accession numbers for the above sequences are DQ455824DQ456480.

Sequence-specific PCR analysis

Real-time PCR was performed to confirm the B. uniformis bloom detected at week 11. Primers 16F2 and 23R4 (Table 1) were used to amplify the rrn ITS from week 11 DNAs and amplicons were cloned into pGEM-T (Kuwahara et al., 2001). Bacteroides uniformis ITS-containing plasmid pBu25 was identified by sequence analysis and subsequently used as the positive control for these experiments. Bacteroides uniformis ITS-specific primers were developed as per the suggestion of Kuwahara et al. (2001): BuQPCRf, BuQPCRr and probe BuFAMprobe (Table 1) (Applied Biosystems) (Kuwahara et al., 2001). This primer/probe set was tested in silico using blast (NCBI). Primers were used at 10 μM with 50 ng of template in 25-μL reactions. Reactions included Platinum Quantitative PCR Supermix-UDG with Rox (Invitrogen), dNTPs (2.5 mM), BSA (500 μg mL−1) and Taq (1.25 U) (Roche). Sequence quantification was performed on an ABI7700 with an initial 10 min of 95°C denaturation followed by 40 cycles of 95°C denaturation (15 s) and 53°C elongation (1 min). Samples were run in triplicate for each of two experiments. Templates were determined to contain different ratios of bacterial to eukaryotic DNAs from sample to sample. Therefore, final quantification was determined by adjusting the number of detected target copies to the percentage of the template that constituted bacterial DNA.

Although no attempt was made to culture Campylobacter from the samples, real-time PCR was performed on the time-course samples to quantify communities of Campylobacter in the birds. Campylobacter coli glyA primers Cc-F1 and Cc-R1 and probe Cc-HEXprobe (Applied Biosystems) were used as described previously except that Cc-HEXprobe was used at a final concentration of 0.3 μM (LaGier et al., 2004).

Sequence-specific dot-blot analyses

Dot blots were performed as described previously (Dore et al., 1998). Total DNA from each animal at 9, 11, 12 and 14 weeks was analysed; in the case of Trial 1, six animals were represented for each time point. In Trial 2, samples were recovered from only four birds during weeks 9 and 12 and five birds from weeks 11 and 14. Then, 250 ng of total DNA from each sample was diluted in 0.1 N NaOH, 1 mM EDTA and applied to positively charged nylon membrane (Amersham) using a Bio-Dot dot blotter (BioRad). Between 4.3 × 1010 and 8.7 × 1010 copies of cloned, purified environmental Clostridium leptum (group IV) and Clostridium coccoides–Eubacterium rectale (group XIV) 16S genes were also applied to the membranes (Collins et al., 1994). Samples were washed with 100 μL of the dilution buffer. 33P-labeled probes Erec482, Clept1240, Euk1379 and Eub338 were diluted in 5 × Denhardt, 5 × SSC, 0.5% sodium dodecyl sulfate (SDS), and salmon sperm DNA (100 μg mL−1 buffer) (Table 1). Membranes were hybridized overnight at 45°C and washed twice for 30 min in 2 × SSC at 47°C (Erec482), 48.5°C (Clept1240), 48°C (Euk1379) and 54°C (Eub338) (Amann et al., 1990; Hicks et al., 1992; Franks et al., 1998; Sghir et al., 2000). Membranes were exposed to a storage phosphor screen for 5 h and the image visualized on a Typhoon 9410 imager (Amersham). Clostridium hybridization is described as the ratio of signal from the specific probe divided by the signal from the universal bacterial probe. Percentage bacterial DNA isolated from cecal samples was calculated from the sum of the hybridization intensities of universal probes Eub338 and Euk1379, under the assumptions that archaebacterial DNA in these samples would be negligible and the universal probes together would detect all or most ribosomal SSU genes.


Species richness and diversity

ARISA analysis of Trial 1 total bacterial community development indicated that species richness increased throughout the trial with Trial 1 averages of 58 ARISA peaks at week 9 and 78 peaks at week 18 (P=0.012) (Table 2). Conversely, Trial 2 averages did not indicate significantly increasing richness between weeks 1 and 18.

View this table:

Weekly cecal bacterial species richness* for time-course Trials 1 and 2, as determined by ARISA

WeekAverage bacterial richness (SD)
Trial 1Trial 2
1ND81 ± 17.0
2ND117 ± 30.6
3ND96 ± 52.8
4ND133 ± 29.9
5ND93 ± 28.9
6ND113 ± 22.2
7ND114 ± 37.2
8ND90 ± 40.6
958 ± 9.8119 ± 9.1
1064 ± 9.8106 ± 4.9
1169 ± 6.3100 ± 18.9
1277 ± 8.993 ± 51.0
1376 ± 11.8120 ± 21.7
1480 ± 5.973 ± 18.4
1584 ± 14.9114 ± 24.2
1685 ± 8.7125 ± 18.3
1780 ± 11.0114 ± 13.5
1878 ± 11.8126 ± 43.2
  • * Richness is defined as the number of peaks in an ARISA profile.

  • Values are means of two ARISA runs ± SD (n=6 in Trial 1; n=5 in Trial 2).

  • ND, not done.

ARISA profile similarities for each bird at each time point are described by NJ dendrograms (Figs 12, 3). Individual birds are represented by letters and weeks by digits. The Trial 1 dendrogram shows tight clusters of early samples (weeks 9–11) and larger clusters containing samples from multiple weeks late in the experiment (weeks 12–18). This clustering pattern indicates rapid change at early time points followed by stabilization as the birds approach puberty. Notable is a substantial transition period centered around week 12. During Trial 2 the data were less clear-cut but indicated a community change around week 11 (Fig. 3).


Overlaid ARISA traces for the five birds in Trial 2 at (a) 9 weeks, (b) 11 weeks, (c) 12 weeks and (d) 14 weeks. ARISA traces were generated by amplification of environmental ribosomal intergenic spacers with FAM-labeled universal primers. Amplification products were then separated on an ABI3100 sequencer. The y-axis represents relative fluorescence units and the x-axis indicates amplicon length.


Dendrogram of ARISA profile similarities of samples from birds A–F, weeks 9–18, Trial 1. Similarities were determined using the Dice Coefficient and the dendrogram was generated using the NJ algorithm.


Dendrogram of ARISA profile similarities from birds G–K, weeks 1–18, Trial 2. Similarities were determined using the Dice Coefficient and the dendrogram was generated using the NJ algorithm.

Sequence library analysis

Clone libraries of time-course Trial 1 weeks 9, 11, 12 and 14 were compared by sequence analysis of 83, 93, 92 and 90 clones, respectively. Chimeras represented 8.4% (week 9), 5.4% (week 11), 21.7% (week 12) and 7.8% (week 14) of the clones (Ashelford et al., 2005). blast and RDPII sequence analyses verified the observed ARISA community transformation (Table 3). At week 9 the birds hosted 37%Clostridiales with little identity to known species. Thirty percent of the clones were identified as Anaerofilum by the RDPII (89% identity over 1361 nt) while those same sequences were identified by blast as Faecalibacterium prausnitzii with 96% identity over 1365 nt. By week 11, 85% of the clones were Bacteroides and of those 99% had 99% identity to B. uniformis over 1447 nt. In weeks 12 and 14, Bacteroidetes continued to dominate, but with a greater diversity of species including 53% and 39% unidentified bacteroidales, respectively. Megamonas hypermegale (AJ420107, 99% identity, 1479 nt), and Shigella boydii (AY696668, 98% identity, 1090 nt) were also detected at week 14. Twenty-eight percent of week 14 clones had similarity to Prevotella spp., and a diverse assortment of Bacteroides with little sequence identity to known species was present. DOTUR assignment of the sequences into OTU according to 1% sequence dissimilarity identified 57 OTU in week 9 [with Shannon–Weaver diversity index (H′)=3.89, evenness (E)=0.96 and Simpson's diversity index (D)=0.01], 19 OTU in week 11 (H′=1.49, E=0.51, D=0.41), 59 OTU in week 12 (H′=2.94, E=0.72, D=0.01) and 65 OTU in week 14 (H′=3.99, E=0.96, D=0.01) (Table 3). Comparison of the four libraries using libshuff indicated significant differences for all pairwise comparisons except those between weeks 9 and 11 and weeks 12 and 14 (Table 4). Community members represented in week 11 were a subset of the week 9 community (P=0.16). Likewise, community members represented in week 14 were a subset of the week 12 community (P=0.095).

View this table:

Predominant bacterial species represented in clone libraries of time-course Trial 1 and Trial 2, weeks 9, 11, 12 and 14

TaxaNumber of clones*
Trial 1Trial 2
Mucispirillum schaedleri1
Bacteroides uniformis57443662913
Total no. of clones7588728388847782
No. of OTU (1% difference)5719596545213443
Shannon–Weaver Index (H′)3.891.492.943.993.311.52.693.06
Evenness (E)§0.960.510.720.960.870.490.760.81
Simpson Index (D)0.010.410.
  • * Numbers indicate total number of clones detected at each time.

  • OTU, operational taxonomic units, where sequences with ≥99% nucleotide identity are considered an OTU.

  • Shannon–Weaver index (H′). This index weights species richness heavily and expressed as H′=−∑pi ln pi, where pi is the proportion of individuals found in the ith OTU. Higher values indicate greater diversity, with 0=complete homogeneity.

  • § Evenness (E). This index reflects differences in relative abundances and is expressed as E=H′/ln S, where H′ is the Shannon–Weaver index and S is the number of OTU. Higher values indicate greater diversity, with 0=complete homogeneity.

  • Simpson Index (D). This index is sensitive to evenness and is expressed as D=∑ni(ni−1)/N(N−1), where ni is the number of individuals for each species and N is the total number of organisms. Higher values indicate lower diversity, with 0=complete heterogeneity.

View this table:

LIBSHUFF comparisons of sequence libraries from weeks 9, 11, 12 and 14 for time course Trial 1 and Trial 2*

XY P-value at week:YX P-value at week:
Trial 1 9Trial 1 11Trial 1 12Trial 1 14
Trial 1 910.1600.0010.001
Trial 1 110.00110.0010.001
Trial 1 120.0010.00110.095
Trial 1 140.0010.0010.0011
Trial 2 9Trial 2 11Trial 2 12Trial 2 14
Trial 2 910.0010.0010.001
Trial 2 110.00110.0010.001
Trial 2 120.0010.92310.116
Trial 2 140.0120.6190.4641
  • * P values<0.05 indicate significant differences between the libraries. In every case, the earlier of the two compared weeks is library X while the latter of the two is library Y. Values below the diagonal reflect XY comparisons and those above the diagonal reflect YX comparisons. If the XY comparison is significantly different but the YX comparison is not significantly different, then X fully represents Y.

Trial 2 chimeras were detected as 5.4% of 93 clones (week 9 library), 9.8% of 92 clones (week 11), 16.3% of 92 clones (week 12) and 5.7% of 87 clones (week 14). blast and RDPII analyses of the week 9 library indicated all clones fell into the Firmicutes with a composition including Faecalibacterium (AJ413954, 96% identity, 1365 bp), Ruminococcus (AJ315979, 96%, 1408 bp) and Subdoligranulum variabile (AJ518869, 97%, 1428 bp) (Table 3). As with Trial 1, the Trial 2 week 11 library was dominated by Bacteroides, 96% of which demonstrated 99% identity over >1400 bp to B. uniformis. Weeks 12 and 14 were composed primarily of Bacteroides and Clostridiales with large populations of B. uniformis (38% and 16%, respectively). Week 14 Bacteroides included Bacteroides vulgatus (M58762, 98% identity over 1451 nt), Bacteroides thetaiotaomicron (AE015928, 96%, 1452 nt) and Alistipes massiliensis (AY547271, 96%, 1450 nt). Clostridiales were the principal members of the week 14 library, and included 43%Faecalibacterium. Eubacterium desmolans (L34618, 97% over 1237 nt) was also represented. Identified week 14 proteobacteria included Escherichia coli (AP009048, 99% over 1465 nt) and Ralstonia pickettii (AY741342, 99% over 1461 nt).

Assignment of clones into OTU was performed using DOTUR (Schloss & Handelsman, 2005). DOTUR assignment identified 45 OTU in the week 9 library (H′=3.31, E=0.87, D=0.05), 21 OTU in week 11 (H′=1.50, E=0.49, D=0.48), 34 in week 12 (H′=2.69, E=0.76, D=0.14) and 43 OTU in week 14 (H′=3.06, E=0.81, D=0.05) (Table 3). libshuff comparisons of the Trial 2 libraries against one another indicated that week 11 was a subset of weeks 12 and 14 (P=0.923 and P=0.619, respectively) and weeks 12 and 14 were not significantly different (P=0.464 for the XY comparison and P=0.116 for the YX comparison) (Table 4).

Significant differences were detected by libshuff between all equivalent time points of the two time courses (data not shown). Bacteroides uniformis subspecies were identified as the cause of the significant difference between the Trial 1 and Trial 2 week 11 libraries. NJ, parsimony and maximum-likelihood analyses all grouped the Trial 1 and Trial 2 B. uniformis 16S gene sequences separately. Confirming the Bacteroides differences, DOTUR assigned the Trial 1 and Trial 2 sequences into different OTU at the 0.1%, but not 1%, sequence dissimilarity level.

Real-time PCR

Real-time PCR quantification of B. uniformis ITS in the cecal dropping DNA of Trial 1 and Trial 2, weeks 9, 11, 12 and 14, was performed in triplicate. Results were adjusted to reflect the copy numbers of B. uniformis ITS sequences per nanogram of bacterial template (Table 5). Real-time primers and probe described in this work gave PCR efficiency >94% and R2>0.98. During Trial 1, B. uniformis increased about twofold from week 9 (9.2 × 105 copies ng−1 DNA) to week 11 (2.3 × 106 copies ng−1) followed by significant decreases in weeks 12 and 14 (1.7 × 105 and 3.5 × 102 copies ng−1, respectively). The trend in Trial 2 included a sixfold increase between weeks 9 and 11 from 1.3 × 104 ITS copies ng−1 to 7.7 × 105 copies ng−1. Bacteroides uniformis copies then subsequently were detected at 6.6 × 105 and 3.3 × 105 copies ng−1 for weeks 12 and 14, respectively. Bacteroides uniformis copy number for Trial 1, week 9 (five B. uniformis library clones), was similar to the copy number for Trial 2 at week 11 (66 clones), indicating that the B. uniformis trend apparent in the clone libraries is probably illusory.

View this table:

Bacteroides uniformis real-time PCR quantification of rrn intergenic spacers (ITS)

B. uniformis ITS copies per ng total DNA (SD)Percentage of total DNA contributed by bacteriaB. uniformis ITS copies per ng bacterial DNA
Trial 1, week 92.7 × 105 (3.8 × 104)309.2 × 105
Trial 1, week 116.0 × 105 (3.0 × 104)272.3 × 106
Trial 1, week 124.6 × 104 (1.3 × 104)271.7 × 105
Trial 1, week 141.1 × 102 (1.2 × 102)323.5 × 102
Trial 2, week 94.52 × 103 (1.4 × 103)361.3 × 104
Trial 2, week 115.3 × 105 (8.6 × 104)687.7 × 105
Trial 2, week 122.8 × 105 (1.1 × 105)426.6 × 105
Trial 2, week 141.4 × 105 (6.4 × 104)443.3 × 105

Campylobacter coli quantification was performed on individual samples of Trial 1 (Table 6). PCR efficiencies were greater than 90% and R2>0.997. Campylobacter quantities during weeks 9 and 11 were low, with averages of 3.6 × 103 and undetectable glyA copies per gram of feces, respectively. Thereafter, copy number increased to an average of 3.3 × 107.

View this table:

Campylobacter coli glyA gene real-time PCR quantification for samples from Trial 1

WeekCopies of glyA per gram feces
Bird ABird BBird CBird DBird EBird F
9NDNDND1 × 1041.2 × 104ND
123.6 × 1072.0 × 1075.0 × 1071.2 × 1086.7 × 1072.2 × 108
144.6 × 1079.1 × 1074.7 × 1067.7 × 1064.1 × 1071.9 × 106
  • ND, not detected.

Dot-blot analyses

Results of the dot-blot hybridizations are given in Table 7. Probe Clept1240, specific for the Clostridium leptum group, generated hybridization ratios of 44% and 52% of the bacterial DNA for weeks 9 of Trial 1 and Trial 2, respectively. Thereafter, for Trial 1, the hybridization ratios dropped to 29%, 20% and 23% for weeks 11, 12 and 14, respectively. During Trial 2 ratios dropped significantly to 13%, 15% and 28% for weeks 11, 12 and 14, respectively. Probe Erec482, specific for the Clostridium coccoides–Eubacterium rectale group, showed a similar pattern with decreases during Trial 1 of 30%, 26%, 27% and 10% signal for weeks 9, 11, 12 and 14, respectively. Trial 2 ratios dropped significantly from 29% to 18%, 15% and 15% during weeks 9, 11, 12 and 14, respectively. All quantities reported represent the percentage of the measured group as compared with the bacterial DNA in each sample. Taken together, these ratios indicate a decrease in the ratio of Clostridium species in the animals over time.

View this table:

Dot-blot hybridization results for Clostridium leptum (group IV) and Clostridium coccoides–Eubacterium rectale (group XIV)

TrialWeekAverage (St. dev) IVXIVTotal percentage
Trial 1943.8 (12.5)29.8 (13.8)73.6
1129.0 (9.5)26.0 (7.0)54.9
1219.9 (13.7)27.4 (14.0)47.4
1422.8 (20.1)9.7 (4.4)32.5
Trial 2951.5 (11.4)29.4 (6.6)80.9
1113.0 (5.8)18.7 (1.2)31.6
1214.9 (3.0)15.4 (3.3)30.3
1428.2 (11.2)14.6 (1.7)42.8


The development of culture-independent methods for the analysis of microbial communities has not only allowed the description of previously unknown organisms but the methods can also be used to decipher the functionality of microbial consortia. Intestinal communities stimulate mucosal angiogenesis and interact with the mucosal epithelium to prevent inflammation while concurrently protecting against pathogens (Stappenbeck et al., 2002). Recent work with the bacterium B. thetaiotaomicron indicates this organism can control the structure of the mucin secreted by Goblet cells and so constrain the composition of the intestinal community (Hooper et al., 2002). An examination of intestinal community development in mice from weaning to puberty identified two transitions, one at weaning and one around day 30 that the authors suggest may have been caused by morphological and immunological maturation of the gut (Inoue et al., 2005). Lacking a weaning-like period, poultry do experience microbiotal maturation over the first weeks of life, including a switch from high to low levels of Lactobacilli between days 3 and 7 (Lu et al., 2003). In the current study, ARISA was used to visualize community maturation in the ceca of male turkeys. Cecal communities from individual animals were followed to identify variability between cohabiting individuals. Previous reports indicate that even very young, cohabiting animals have unique intestinal communities probably derived from variable immune responses and the order in which the intestinal habitat is exposed to individual microbial species (van der Wielen et al., 2002).

ARISA patterns in this study were interpreted as descriptive of microbial composition, with the understanding that a peak may contain amplicons from multiple species and a single species may contribute multiple and differently sized amplicons. Additionally, only species representing >1% of the community may contribute detectable amplicons (Muyzer et al., 1993). In the intestine many closely related species have evolved from a common ancestor within their animal hosts (Backhed et al., 2005). Consequently, these species may not be discriminated well by ARISA, leading to diversity underestimation. Clustering of early (preceding week 12) ARISA profiles by week, and not by animal, indicates high community similarity between individual birds and parallel microbial maturation patterns between individual animals in Trial 1. Variability between individual birds can be seen in Fig. 2. For example, the bacterial community of bird F was different from those of its cohorts from weeks 10 to 14, and this community appeared to stabilize before those of the other birds. Clustering of late (weeks 12–18) ARISA profiles by animal rather than week indicates stabilization of the bacterial communities, although it is unclear whether climax assemblages were established. These results probably reflect both limited exposure to microorganisms in the containment barn and inbreeding associated with commercial turkey flocks (Zoetendal et al., 2001).

The community transitions observed around week 12 have not previously been reported. The shifts were described in detail by sequence analysis of SSU clones amplified from weeks 9, 11, 12 and 14. The week 9 libraries were dominated by Clostridiales, followed by a large transformation to principally Bacteroidetes. Faecalibacterium prausnitzii predominated at week 9 in both trials. This microorganism is a member of the Cl. leptum subgroup IV and a predominant microorganism in the intestines of many mammals as well as poultry (Wang et al., 1996; Gong et al., 2002). It is ureolytic, has a requirement for acetate, and produces butyrate, formate and lactate (Duncan et al., 2002). Reductions in carriage of this microorganism have been correlated with frailty in older adult humans (van Tongeren et al., 2005). At week 11, 84 and 79% (Trial 1 and Trial 2, respectively) of the clones had high identity to B. uniformis, an organism known to deconjugate bile acids and respond positively to Di-d-fructofuranose-1,2′ : 2,3′-dianhydride (DFA III) in the rat intestine, correlating with decreased pH and increased short-chain fatty acids (Shindo & Fukushima, 1976; Minamida et al., 2005). The B. uniformis strain IK was identified as acting synergistically with Treponema hyodysenteriae to produce intestinal lesions in mice (Hayashi et al., 1990). In addition, B. uniformis stimulation of transforming growth factor-β1 (TGF-β1) has been associated with deregulation of the intestinal mucosa repair functions (Border & Ruoslahti, 1992; Mourelle et al., 1998). Real-time PCR quantification of B. uniformis did not confirm a dominant role for the microorganism in the transition, but nevertheless it was established as a plentiful member of the microbiota.

Either positive or negative influences could cause the community transformations described here. The impact of the diet regimen is one possible explanation. Birds were raised from hatching to 6 weeks on a 28% protein diet and then switched to a 22% protein diet containing Lactobacillus, Bifidobacterium and Enterococcus fermentation products as described on the manufacturer's label. Energy consumption by male turkeys decreases to <1 Mcal kg−1 body weight around week 12 and the protein requirement decreases. Under commercial practices the diet would be switched again to 16% protein at week 12, but in the current work birds were maintained on the 22% diet in an effort to minimize variability in the trials. The shift could be a result of excess dietary protein ingested by the older animals, or related to the changing nutritional needs of the animals. Dietary effects on intestinal communities are strongly supported, but neither the timing of community structure rearrangements nor the extent of the rearrangements is well documented (Hentges, 1980; Rowland et al., 1985; Mai, 2004). Other positive functions such as host-derived signals may stimulate the transformation. Possible host-derived signals include puberty, although measurable levels of testosterone do not accumulate until week 19 (Bacon et al., 2000). Other signals could be derived from the expansion of the gut-associated lymphoid tissue that occurs between 6 and 12 weeks of age and the concurrent increase in TCR1-expressing T cells between 9 and 12 weeks (Befus et al., 1980; Lillehoj & Chung, 1992). Finally, week 11 is the time at which growth hormone (GH) falls precipitously from juvenile levels (>85 ng mL−1 in plasma) to adult levels (<45 ng mL−1 in plasma; Scanes et al., 1984). GH stimulates the thyroid, adrenal glands and pancreas and affects lipid and sugar metabolism. Some negative functions imposed by the host could include decreased intestinal pH or increased defensin production.

Weeks 11 and 12 were a time of great flux and we hypothesize that these periods may be a time of increased susceptibility to colonization by food-borne pathogens (Backhed et al., 2005). If cecal microbial communities are stable, all physical and physiological niches should be filled. However, if cecal microbial communities are immature or in flux, the possibility remains for opportunistic colonization by pathogens (Lee, 1985; Bailey & Coe, 1999). Real-time quantification of Ca. coli in Trial 1 indicates a bloom of the pathogen at week 12. Whether this correlation supports our hypothesis is unknown as this may be the first report quantifying Campylobacter throughout turkey growout. Campylobacter was not detected during Trial 2, so real-time data were not gathered.

The two primary purposes of ecological analysis are description of the biota in an ecosystem and functional analysis of populations. In this work, comprehensive (ARISA) followed by detailed (sequence library) description of the turkey cecal microbiota has identified a potentially important host–microorganism interaction with implications for human and animal health, nutrition, and food safety. Initial communities appear to develop sequentially, but around weeks 11 and 12 a substantial transition was observed. This transformation may derive from a host signal such as a reduced need for protein or decreased growth hormone, and further studies along these lines are warranted. The detection of Ca. coli correlated with the transformation suggests this period may provide an opportunity for colonization of the intestine by food-borne pathogens and may be a critical control point for exclusion of Campylobacter from poultry.


I gratefully acknowledge Jennifer A. Jones for generation of the sequence libraries and sequence alignment, and Dr David P. Alt and Karen Halloum for operation of the ABI3100 for both sequence and ARISA analyses.


  • Editor: Julian Marchesi


View Abstract