RESEARCH ARTICLE
O al Samples as Non-In asi e P oxies o
Assessing he Composi ion o he Rumen
Mic obial Communi y
Ilma Tapio
1
, Ke in J. Shing ield
1¤
, Nes McKain
2
, Au élie Bonin
3
, Daniel Fische
1
, Ali
R. Baya
1
, Johanna Vilkki
1
, Pie e Tabe le
3
, Timo hy J. Snelling
2
, R. John Wallace
2
*
1G een Technology, Na u al Resou ces Ins i u e Finland, Jokioinen, Finland, 2Rowe Ins i u e o Nu i ion
and Heal h, Uni e si y o Abe deen, Abe deen, Uni ed Kingdom, 3Labo a oi e d'Ecologie Alpine, CNRS,
G enoble, F ance
¤Cu en add ess: Ins i u e o Biological, En i onmen al and Ru al Sciences, Abe ys wy h Uni e si y,
Abe ys wy h, Uni ed Kingdom
*[email p o ec ed].uk
Abs ac
Mic obial communi y analysis was ca ied ou on uminal diges a ob ained di ec ly ia
umen is ula and buccal luid, egu gi a ed diges a (bolus) and aeces o dai y ca le o
assess i non-in asi e samples could be used as p oxies o uminal diges a. Samples we e
collec ed om i e cows ecei ing g ass silage based die s con aining no addi ional lipid o
ou di e en lipid supplemen s in a 5 x 5 La in squa e design. Ex ac ed DNA was analysed
by qPCR and by sequencing 16S and 18S RNA genes o he ungal ITS1 amplicons. Fae-
ces con ained ew p o ozoa, and bac e ial, ungal and a chaeal communi ies we e subs an-
ially di e en o uminal diges a. Buccal and bolus samples ga e much mo e simila
p o iles o uminal diges a, al hough ewe a chaea we e de ec ed in buccal and bolus sam-
ples. Bolus samples o e all we e mos simila o uminal samples. The di e ences be ween
bo h buccal and bolus samples and uminal diges a we e consis en ac oss all ea men s. I
can be concluded ha ei he p oxy sample ype could be used as a p edic o o he umen
mic obial communi y, he eby enabling mo e con enien la ge-scale animal sampling o
pheno yping and possible use in u u e animal b eeding p og ams aimed a selec ing ca le
wi h a lowe en i onmen al oo p in .
In oduc ion
Ruminan li es ock p oduc ion has a la ge en i onmen al oo p in because o emissions o he
g eenhouse gas, me hane, and o high ni ogenous emissions in u ine and aeces [1]. The p i-
ma y sou ce o hese emissions is umen mic obial me abolism [2]. Thus, unde s anding he
uminal mic obio a is a i al p e equisi e o imp o ing he en i onmen al c eden ials o mea
and milk p oduc ion. The in oduc ion o high- h oughpu sequencing echniques has opened
new ways o explo e complex mic obial ecosys ems, including he umen [3–6]. Sampling
uminal diges a is essen ial o enable he echnology bene i s o be ealized. Rumen sampling
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 1/15
OPEN ACCESS
Ci a ion: Tapio I, Shing ield KJ, McKain N, Bonin A,
Fische D, Baya AR, e al. (2016) O al Samples as
Non-In asi e P oxies o Assessing he Composi ion
o he Rumen Mic obial Communi y. PLoS ONE 11
(3): e0151220. doi:10.1371/jou nal.pone.0151220
Edi o : Robe J Fo s e , Ag icul u e and Ag i-Food
Canada, CANADA
Recei ed: Sep embe 21, 2015
Accep ed: Feb ua y 23, 2016
Published: Ma ch 17, 2016
Copy igh : © 2016 Tapio e al. This is an open
access a icle dis ibu ed unde he e ms o he
C ea i e Commons A ibu ion License, which pe mi s
un es ic ed use, dis ibu ion, and ep oduc ion in any
medium, p o ided he o iginal au ho and sou ce a e
c edi ed.
Da a A ailabili y S a emen : Da a a e a ailable om
he D yad da abase (accession numbe h p://dx.doi.
o g/10.5061/d yad.1b07d).
Funding: This wo k was suppo ed by he Eu opean
Commission 289319.
Compe ing In e es s: The au ho s ha e decla ed
ha no compe ing in e es s exis .
can be ca ied ou by o al in uba ion, bu his is an unpleasan p ocedu e o he animal and
also esul s in a sample ha is o en hea ily con amina ed wi h sali a [7]. Rumenocen esis p o-
ides alid samples bu in ol es punc u ing he abdominal wall wi h a needle and emo ing
diges a by sy inge, also undesi able in e ms o animal wel a e [8], and es ic s he amoun o
sample ha can be collec ed. The mos eliable samples a e ob ained om animals ha ha e
been su gically modi ied by i ing a uminal cannula [7,9], bu his equi es skilled su ge y,
dedica ed animal acili ies and in mos coun ies equi es o mal go e nmen al pe mission. In
any case, umen is ula ion is imp ac ical o sampling la ge numbe s o animals.
Ruminan s egu gi a e uminal con en s egula ly in o de o chew he pa ially diges ed
plan ma e ial [10,11]. The chewed bolus is hen swallowed o u he mic obial deg ada ion.
I migh be expec ed, he e o e, ha he mic obiome o he mou h could ep esen a e lec ion
o he uminal mic obiome. I so, collec ion o small samples o o al luid could be used as a
p oxy o assessing he mic obial ecology o he umen, a oiding he need o mo e in asi e
sampling p ocedu es. Ou hypo hesis was ha he o al sample will con ain mic obes om he
egu gi a ed bolus ha esul s om umina ion and ha he mic obial composi ion o he
bolus migh be ep esen a i e o he uminal communi y. Ano he p oxy ha has been in es i-
ga ed be o e is aeces. The aecal communi y is signi ican ly di e en o ha o he umen [12–
15], bu none heless he e may be indica o s ha could p o e use ul, analogous o he p esence
o aecal a chaeol, a memb ane lipid o uminal a chaea, being used as a ma ke o uminal
me hanogenesis [16–18]. The aim o he p esen expe imen was o compa e he communi ies
in hese po en ial al e na i e samples in o de o e alua e hei use ulness as p oxies o di ec
sampling o uminal diges a. A pape has been published ecen ly [19] in which he same
hypo hesis was explo ed using buccal samples in sheep. The p esen pape con i ms he con-
clusions o ha pape conce ning he alidi y o buccal sampling, in his case using dai y cows,
and u he ampli ies he in es iga ion by compa ing bolus and aecal samples.
Ma e ials and Me hods
Animal expe imen a ion
All expe imen al p ocedu es we e app o ed by he Na ional E hics Commi ee (Hämeenlinna,
Finland) in acco dance wi h he guidelines es ablished by he Eu opean Communi y Council
Di ec i es 86/609/EEC [20]. The expe imen was conduc ed be ween 15 Feb ua y 2012 and 3
July 2012 a Na u al Resou ces Ins i u e Finland ( o me ly MTT Ag i ood Resea ch Finland),
Jokionen Finland (60.8° N, 23.5° E; al i ude 103 m). Fi e Finnish Ay shi e cows i ed wi h
umen cannula (#1C, i.d. 100 mm, Ba Diamond, Inc., Pa ma, ID) o (mean ± SE) 4 ± 0.6 pa -
i y, 63 ± 11.7 d in milk, and 705 ± 25.5 kg li e weigh we e used in a 5 × 5 La in squa e wi h
28-d expe imen al pe iods. T ea men s comp ised o al mixed a ions based on g ass silage
( o age: concen a e a io 60:40 on a d y ma e (DM) basis) con aining no addi ional a (CO)
o 50 g/kg die DM o me hyl es e s o my is ic acid (MA), apeseed oil (RO), sa lowe oil
(SO) o linseed oil (LO). Lipid supplemen s eplaced concen a e ing edien s. Each pe iod
comp ised 2 d adap a ion, 21 d supplemen a ion, and 5 d washou o minimize ea men
ca y-o e e ec s. Samples o he analysis o umen mic obial composi ion we e collec ed a
15.00 h on d 20 and 09.00 h on d 22 o each pe iod.
Collec ion and p ocessing o samples
Ruminal diges a samples we e collec ed om ou egions (an e io do sal, an e io en al,
pos e io do sal, and pos e io en al) wi hin he umen- e iculum. Immedia ely a e collec-
ion, uminal diges a samples we e mixed ho oughly and squeezed h ough 2 laye s o
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 2/15
cheeseclo h. Fi e hund ed μl o umen liquid we e mixed wi h 1 ml o phospha e bu e ed
saline-glyce ol (30% / ) bu e (PBS-gly) and immedia ely ozen a -80°C.
Regu gi a ed inges a (bolus) samples we e collec ed as close in ime as possible o ha o
umen samples. Depending on he umina ion beha iou o each cow, he ime a e umen
sampling a ied among he animals bu did no exceed 20 min. Bolus samples we e p ocessed
in he same way as umen samples.
Buccal samples (e ec i ely sali a mixed wi h bolus pa icles) we e collec ed a he same
ime as bolus using sponge swabs. Th ee collec ion me hods we e in es iga ed: (i) samples
we e collec ed using he BuccalAmp DNA ex ac ion ki (Epicen e) and (ii) using he Pe o -
magene Li es ock (PG-100) ki (DNA geno ek) and p ocessed ollowing he manu ac u e ’s
p o ocols, o (iii) samples we e collec ed using Pe o magene Li es ock sponge swabs, immedi-
a ely subme ged in 1 ml o PBS-gly bu e and ozen a -80°C. P ocessing o samples based on
me hod (iii) was he bes in e ms o DNA quali y and quan i y, and he e o e used o sam-
pling. Fo echnical easons, buccal swabs could no be aken du ing he i s expe imen al
pe iod.
F esh aeces we e collec ed by s imula ing ec al ac i i y a he ime a ound umen sam-
pling. Twen y i e g o aeces we e p ese ed in 50 ml o PBS-gly bu e and s o ed a -80°C.
To al genomic DNA was ex ac ed om 1 ml o mixed d 20 and d 22 sample (in case o
umen, bolus and buccal swab) o 30 mg o aeces ollowing he p o ocol o Yu and Mo ison
[21].
Quan i a i e PCR o 16S and 18S RNA genes
DNA concen a ions we e de e mined wi h a NanoD op ND 1000 Spec opho ome e (Nano-
D op Technologies, Wilming on, DE). DNA was dilu ed o 0.1 ng/μlin5μg/ml he ing spe m
DNA o ampli ica ion wi h uni e sal bac e ial p ime s UniF and UniR [22] and 1 ng/μlin
5μg/ml he ing spe m DNA o ampli ica ion o o he g oups [23]. Quan i a i e PCR was ca -
ied ou using a BioRad CFX96 as desc ibed by Rami ez-Fa ias e al. [24]. P ime se s and a -
ge species a e desc ibed in Table A in S1 Tex . Ampli ica ion o a chaeal 16S RNA genes was
ca ied ou using he p ime s desc ibed by Hook e al. [25] and calib a ed using DNA ex ac ed
om Me hanob e ibac e smi hii PS, a gi om M. P. B yan , Uni e si y o Illinois. Ampli ica-
ion e iciency was e alua ed using empla e DNA om Rosebu ia hominis A2-183 (DSM
16839
T
) o he uni e sal bac e ia and Clos idium Clus e XIVa calib a ions, Faecalibac e ium
p auzni zii A2-165 (DSM 17677
T
) o Clos idium Clus e IV, and Bac e oides he aio aomi-
c on VPI-5482 (DSM 2079
T
) o Bac e oide es. P o ozoal 18S RNA gene ampli ica ion was
calib a ed using DNA ampli ied om bo ine umen diges a wi h p ime s 54 and 1747 [26].
Co e age o qPCR p ime s was checked om o iginal e e ences and by use o he P obe
Ma ch ool o he Ribosome Da abase P ojec [27]. Bac e ial abundance was calcula ed om
quad uplica e C alues using he uni e sal bac e ial calib a ion equa ion. Pai wise di e ences
be ween ea men s and be ween samples wi hin ea men s we e e alua ed by a simple - es .
P alues ha e been adjus ed o mul iple es ing using he Benjamini-Hochbe g me hod [28].
Amplicon p epa a ion and sequencing
P ime s used o PCR ampli ica ion o bac e ia and a chaea 16S RNA genes, cilia e p o ozoa
18S RNA genes and anae obic ungi ITS1 genes we e designed in silico using ecoP ime s [29],
he OBITools so wa e sui e [30](h p://www.g enoble.p abi. / ac/OBITools) and a da abase
c ea ed om sequences s o ed in GenBank. Fo each sample, PCR ampli ica ions we e pe -
o med in duplica e. An eigh nucleo ide ag unique o each PCR duplica e was a ached o he
p ime sequence, in o de o enable he pooling o all PCR p oduc s o sequencing and he
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 3/15
subsequen assigna ion o sequence eads o hei espec i e samples. PCR amplicons we e
combined in equal olumes and pu i ied (QIAquick PCR pu i ica ion ki , Qiagen, Ge many).
Amplicon lib a ies we e p epa ed in Fas e is SA (Gene a, Swi ze land h ps://www. as e is.
com/dna/) using he T uSeq Nano DNA HT Sample P ep Ki om Illumina using a p o ocol
wi h only i e PCR cycles (h ps://www. as e is.com/dna/?q=con en /me a as -p o ocol-
amplicon-me agenomic-analysis). All ma ke s we e sequenced using he MiSeq echnology
om Illumina, which p oduced 250-base pai ed-end eads, excep o he a chaea ma ke ,
which was sequenced on he Illumina HiSeq pla o m, gene a ing 100-base pai ed-end eads.
In silico analysis showed ha sho e a chaea amplicons did no comp omise abili y o iden i y
a chaea a species le el.
Sequence analysis and axonomic assignmen
Alignmen o pai ed-end eads, sample assignmen and emo al o sequences wi h ambiguous
nucleo ides and sequences o leng hs ou side he empi ical sequence leng h dis ibu ion we e
pe o med wi h he OBITools so wa e sui e. Sequences we e deposi ed in D yad da abase
unde accession numbe h p://dx.doi.o g/10.5061/d yad.1b07d. Sequences we e clus e ed in o
ope a ional axonomic uni s (OTU) a 97% simila i y using UCLUST [31] and il e ed o chi-
me ic eads using Chime aSlaye (bac e ia, p o ozoa and a chaea) o UCHIME ( ungi) as
implemen ed in QIIME pipeline 1.7.0 [32]. Taxonomy was assigned using he BLAST me hod
[33]. Bac e ial OTUs axonomy was assigned using he G eengenes 12_10, a chaeal—RIM-DB
da abase [34], ha o cilia e p o ozoa using he SILVA 18S da abase [35] and anae obic ungi
we e assigned using a cu a ed ungal ITS e e ence da abase [36] kindly p o ided by AgRe-
sea ch L d (Palme s on No h, New Zealand). Single on OTUs we e emo ed and he da a
om each sample we e a e ied o he simila sequencing dep h p io o u he analyses using
QIIME. Pai wise axonomy compa isons we e pe o med by compu ing Pea son co ela ion
coe icien s as implemen ed in QIIME. In o al, 20 umen-bolus-buccal swab-and aecal sam-
ples collec ed om he same animals a he same ime we e compa ed. Sca e plo analysis was
done using R 2.15.0 [37].
Fo c ea ing mic obial co-occu ence ne wo ks in umen and al e na i e sampling si es, he
Spa CC mic obial associa ion ne wo k in e ence ool [38] was used o calcula e co ela ion
coe icien s be ween all bac e ia, a chaea, cilia e p o ozoa and ungi a he genus o he deepes
iden i iable axonomic classi ica ion le el. Co ela ions we e de i ed whe e X = (x_ij), i,
j = 1,2,...,m is he esul ing co ela ion ma ix wi h x_ij being he pai wise co ela ion be ween
mic obes i and j and m being he o al amoun o compa ed mic oo ganisms. An adjacency
ma ix A = (a_ij), i,j = 1,2,...,m was de e mined whe e a_ij = 1, i x_ij>= 0.25, a_ij = -1, i
x_ij <= -0.25 and a_ij = 0 else. Once de e mined, he adjacency ma ix was used o cons uc a
co-occu ence ne wo k, whe e each node ep esen s a axon while he edges be ween he nodes
ep esen posi i e/nega i e co ela ions be ween axa. Communi ies wi hin he ne wo ks we e
iden i ied by applying he leading eigen ec o [39] me hod using he R-Package 'ig aph' [40].
The pai wise simila i ies be ween he communi y s uc u es o ne wo ks ha e been e alua ed
o each ea men and sampling si e sepa a ely using he adjus ed and index [41] and isual-
ized as a hea map, whe e b igh e colou s e e o la ge and alues indica ing close
simila i y.
Resul s
In o al, 80 samples, collec ed om ou sampling si es, he umen, buccal luid, bolus and ae-
ces we e compa ed. Samples we e collec ed om 5 lac a ing cows used in a 5 × 5 La in Squa e
wi h 28-d expe imen al pe iods o e alua e 5 expe imen al die s. T ea men s comp ised o al
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 4/15
mixed a ions based on g ass silage con aining no addi ional a (CO) o supplemen ed wi h
me hyl my is a e (MA), apeseed oil (RO), sa lowe oil (SO) o linseed oil (LO). Die s we e
o mula ed o induce changes in umen mic obial popula ions o p o ide a obus es o he
sui abili y o sampling p oxies.
qPCR analysis o 16S and 18S RNA genes om uminal, buccal, bolus
and aecal samples
The abundance o di e en mic obial g oups was compa ed by qPCR ac oss ea men s and
sample ypes (Fig 1; Tables B and C in S1 Tex ). Faecal samples di e ed ma kedly om co e-
sponding uminal samples, in ha al hough a chaea we e p esen a a simila abundance, p o-
ozoa we e i ually absen , and o al bac e ia we e highe . Clos idium Clus e IV was on
a e age 5× mo e abundan in aeces han in uminal diges a. Bac e oide es showed a co e-
spondingly lowe abundance.
Buccal swab samples con ained simila abundances o o al bac e ia o uminal samples, bu
he p opo ion o Bac e oide es ended o be highe and Clos idium Clus e s IV and XIVa
lowe . A chaea we e 0.48× as abundan in swab samples compa ed o uminal samples. P o o-
zoal 18S RNA abundance appea ed o be 1.7× highe in swab samples.
Samples om he bolus o egu gi a ed diges a we e gene ally close in p o ile o uminal
diges a samples emo ed ia he uminal is ulae. A chaea we e 0.80×, p o ozoa we e 0.75×
and o al bac e ia we e 1.22× he abundance in uminal diges a, al hough di e ences we e no
s a is ically signi ican due o high a iabili y be ween samples. The p opo ions o he di e en
classes o bac e ia we e simila .
No di e ences (FDR <0.05) we e de ec ed due o ea men in any o he sample ypes,
excep ha cilia e p o ozoa ended o be dec eased in umen and o al samples by die a y MA
supplemen s, wi h a compensa o y inc ease in o al bac e ia.
Mic obial composi ion o di e en sampling si es by RNA gene amplicon
analysis
Twen y umen-bolus-buccal swab- aeces samples se s we e collec ed o sampling si e compa -
isons. In o al, 7,305,504 high quali y sequencing eads ac oss all 4 majo mic obial g oups
(bac e ia, a chaea, cilia e p o ozoa and anae obic ungi) we e gene a ed. The numbe o
sequences assigned o each mic obial g oup is epo ed in Table D in S1 Tex . Mic obio a com-
posi ion is p esen ed bo h as ela i e abundance in ex ac ed DNA (Fig 2) and he di e ence in
abundance ela i e o ha de e mined in uminal diges a (Fig 3; Fig A in S1 Tex ).
One hund ed and ou genus-like g oups o bac e ia we e iden i ied in he o al da ase .
Fi y o hese had an a e age abundance abo e 0.5% in a leas one o he ea men s and sam-
pling si es bu only 16 could be iden i ied o he genus le el. The emaining 54 g oups wi h
a e age abundance below 0.5% we e pooled a he lowes common axonomic le el. In o al,
hey accoun ed o less han 5% o all sequencing eads. The a chaeal communi y was ep e-
sen ed by 9 gene a consis ing o 15 g oups a he species le el, while cilia e p o ozoa and ungal
popula ions we e ep esen ed by 14 g oups a he genus le el, espec i ely.
The bac e ial communi ies in uminal, bolus and buccal swab samples we e ema kably
simila o each o he and a ec ed li le by ea men . The phylum Bac e oide es was simila in
umen and bolus samples (ca. 45% o o al sequences) bu was 10% highe in buccal swabs and
20% lowe in aeces. These di e ences we e mainly ela ed o he p opo ion o P e o ella,
which was he mos abundan bac e ial genus in umen-bolus-swab samples (Fig 2). The Rike-
nellaceae amily and he phylum Ac inobac e ia we e de ec ed in aeces bu no in o he sam-
ples. Bolus samples also ep esen ed a close ma ch o umen samples in he abundance o
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 5/15

Fi micu es (35%) while in aecal samples Fi micu es accoun ed o 70% o all sequences, wi h
he Ruminococcus genus being he mos abundan (Fig 2). Ou o eigh dis inc bac e ial g oups
wi hin he Lachnospi aceae amily, he mos ob ious di e ences among he sampling si es we e
ela ed o a g oup o bac e ia classi ied only a amily le el (Lachnospi aceae1). This g oup was
o e ep esen ed in bolus, unde ep esen ed in swabs and absen in aecal samples. Di e ences
be ween umen and swab samples we e ela ed o lowe abundance o Clos idia1,Ruminococ-
caceae1,Co iobac e iales g oups and Bu y i ib io genus in swabs, accoun ing o 1–2.5% lowe
abundances compa ed o uminal diges a (Fig 3a).
In he a chaeal communi ies, Me hanob e ibac e domina ed in umen-bolus-swab samples
bu he p opo ion o Mbb.go schalkii was unde ep esen ed and Mbb. uminan ium o e es i-
ma ed in bolus and buccal swab samples compa ed o uminal samples (Figs 2and 3b). Me ha-
nosphae a was he second mos common genus among a chaea in hese samples. In aeces, he
Me hanoco pusculum genus, no de ec ed in umen-bolus-swab samples, accoun ed o up o
53% o sequencing eads, wi h Me hanob e ibac e and Me hanosphae a gene a being less
abundan . Fi e dis inc g oups wi hin he Me hanomassiliicoccaceae amily we e de ec ed in
umen-bolus-swab samples a simila abundance below 0.1% (Fig 3b). Die had li le in luence
on he a chaeal communi y o he di e en sample ypes (Fig 2).
Cilia e p o ozoa composi ion in bolus samples ma ched closely umen samples bu di e -
ences we e obse ed in cilia e axon abundance. En odinium was mo e abundan whe eas Iso-
icha and Me adinium mino um we e less abundan in bolus compa ed o umen samples.
Buccal swab samples con ained a high p opo ion o Diplodinium (>33%) compa ed o umen
Fig 1. qPCR o 16S and 18S RNA genes in buccal swabs, egu gi a ed diges a (bolus), uminal
diges a and aeces. Samples collec ed om lac a ing cows ed a g ass silage based die con aining no
addi ional a (CO) o supplemen ed wi h 50 g/kg die d y ma e o me hyl my is a e (MA), apeseed oil (RO),
linseed oil (LO) o sa lowe oil (SO). Resul s a e exp essed as copy numbe pe ng o ex ac ed DNA. E o
ba s ep esen SD, n = 4 pe ea men , excep o ou buccal samples, one each om CO, RO, LO and SO,
om which sa is ac o y ampli ica ion was no ob ained.
doi:10.1371/jou nal.pone.0151220.g001
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 6/15
samples (<1%), whe eas he p opo ion o uncul u ed En odinium was lowe by 16–23%. Fu -
he mo e, he ela i e abundance o Me adinium mino um and Raabena bella in buccal swabs
was below 1% compa ed o 6–17% in umen and bolus samples, while Polydiniella myso ea
was de ec ed in swabs (1–2%) bu only in ace amoun s in o he sample ypes. No umen spe-
ci ic p o ozoa we e obse ed in aecal samples. The much lowe abundance o aecal p o ozoa
included he pa asi ic cilia e p o ozoa, Balan idium coli and Blas ocys is sp. ha we e absen
om umen-bolus-swab samples. T ea men had li le in luence on he p o ozoal communi y
o he di e en sample ypes (Fig 2).
Die a y lipid supplemen s had a p o ound e ec on he ungal popula ion composi ion,
pa icula ly me hyl my is a e (Fig 2). Apa om O pinomyces 1b ound only in aeces a an
Fig 2. Rela i e abundance o bac e ia, a chaea, p o ozoa, and ungi based on amplicon sequencing o
16S-18S RNA genes, and ITS1 sequences in umen, bolus, buccal swabs and aeces. Samples
collec ed om lac a ing cows ed o al mixed a ions based on g ass silage con aining no addi ional a (CO),
o supplemen ed wi h 50 g/kg d y ma e o me hyl my is a e (MA), apeseed oil (RO), sa lowe oil (SO) o
linseed oil (LO). Da a epo ed based on he mean o 4 animals pe die a y ea men .
doi:10.1371/jou nal.pone.0151220.g002
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 7/15
abundance below 3%, he composi ion o he ungal popula ion in umen-bolus-swab samples
was simila . Di e ences we e ela ed o a ia ion in abundance o Caecomyces1, KF1, Neocalli-
mas ix1 and SK3 ungal g oups. The closes ma ch was obse ed be ween umen and bolus
samples (Fig 3b).
Co espondence analysis o mic obial composi ion be ween sample
ypes
To e alua e how well he mic obial communi ies in bolus, buccal swabs o aecal samples ep-
esen ed ha in uminal diges a, sca e plo s we e gene a ed (Fig 4). In he sca e plo s he el-
a i e abundances o e e y mic obial axon om each animal, ep esen ing all 3 non-in asi e
sampling si es, we e compa ed wi h umen samples. The s eng h o simila i y be ween sam-
pling si es was es ima ed as an o e all Pea son co ela ion coe icien . The co ela ions de i ed
indica ed ha bolus samples ma ched mos closely he umen composi ion o bac e ia
( alue = 0.99), while li le o e all co espondence was obse ed be ween aecal and umen
samples. E en hough di e ences in abundance we e obse ed among he mo e common bac-
e ial g oups in he buccal swab- umen compa ison, he o e all co ela ion was high none he-
less ( alue = 0.98). Pa e ns o simila i y o a chaea and anae obic ungi among sampling
si es was simila as o bac e ia (Fig 4a and 4b). The la ges di e ence be ween bolus and buccal
samples compa ed wi h umen diges a was o cilia e p o ozoa, while no ela ion o aecal p o-
ozoal composi ion was de ec ed (Fig 4b).
Fig 3. Changes in mic obial abundance be ween umen and he h ee al e na i e sampling si es. Each ow ep esen genus-like mic obial g oup o a)
bac e ia and b) a chaea, cilia e p o ozoa, and ungi, while each do ep esen s indi idual cow. Di e ences we e calcula ed as ela i e abundance in buccal
swab minus abundance in umen ( ed ), aeces— umen (g een ) and bolus— umen (blue ) based on n = 20 pe sampling si e.
doi:10.1371/jou nal.pone.0151220.g003
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 8/15
Mic obial co-occu ence analysis
The po en ial o collec ing bolus, aeces o buccal swabs as a eliable al e na i e o umen sam-
pling was examined h ough he gene a ion o mic obial associa ion ne wo ks. Analysis was
pe o med o each die a y ea men and on each sample ype. Only co ela ions wi h Spa CC
| | 0.25 we e used o cons uc ing he ne wo ks. Mic obial communi ies iden i ied wi hin
he ne wo ks o al e na i e sampling ypes we e compa ed wi h umen communi ies. Simila i-
ies o he ne wo ks a e p esen ed as a hea map (Fig B in S1 Tex ). Mic obial ne wo ks iden i-
ied in bolus and swab samples did no o e an exac ma ch o he ne wo k composi ion
obse ed in umen samples, while aecal mic obial in e ac ions we e no compa able o he
o he sampling si es. Die a y ea men s appea ed o ha e an e ec on he di ec o indi ec
in e ac ions be ween mic obial axa, wi h changes in esponse o MA showing he mos dis inc
di e ences.
Use o a chaeal abundance as a bioma ke o me hane emissions
The qPCR da a o indi idual animals we e used o calcula e he a io o abundance o a chaea
and bac e ia, and he e ec s o die a y ea men and di e en sample ypes we e compa ed.
Inclusion o addi ional lipid in he die a y ea men dec eased he ela i e p opo ions o
a chaea ela i e o bac e ia in umen samples om MA (P= 0.013) and LO (P= 0.033) ea -
men s, wi h he esponses o MA being he mos p onounced, lowe ing he a io o uminal
a chaea:bac e ia abundance by 36% compa ed o he con ol (Table E in S1 Tex ). The e ec s
o lipid supplemen s we e also e iden in he o he sample ypes, bu he a chaeal abundance
was signi ican ly lowe in all o he samples compa ed o uminal diges a (Table E in S1 Tex ).
When indi idual animals we e compa ed ac oss ea men s, a close co ela ion exis ed
Fig 4. Sca e plo analysis o di e ences in ela i e abundance, es ima ed o each mic obial axon in all indi idual animals and ac oss sample
ypes, espec i ely. a) Each bac e ial axon in samples collec ed om all cows is ep esen ed as sepa a e () in he lowe iangle, while a chaea a e
p esen ed as (Δ) in he uppe iangle; b) Anae obic ungi a e ma ked as (□) and cilia e p o ozoa as (●). The a e age Pea son co ela ion coe icien is
indica ed in he bo om- igh co ne o each diag am.
doi:10.1371/jou nal.pone.0151220.g004
P oxies o Rumen Samples
PLOS ONE | DOI:10.1371/jou nal.pone.0151220 Ma ch 17, 2016 9/15