Gene . Sel. E ol. 40 (2008) 195–214 A ailable online a :
c
INRA, EDP Sciences, 2008 www.gse-jou nal.o g
DOI: 10.1051/gse:2007044
O iginal a icle
Quan i a i e ai loci o e ili y ai s
in Finnish Ay shi e ca le
Nina F. Schulman1∗, Gou am Sahana2, Mogens S. Lund2,
Si ja M. Vii ala1, Johanna H. Vilkki1
1MTT Ag i ood Resea ch Finland, Bio echnology and Food Resea ch,
31600 Jokioinen, Finland
2Depa men o Gene ics and Bio echnology, Facul y o Ag icul u al Science,
Aa hus Uni e si y, Resea ch Cen e Foulum, 8830 Tjele, Denma k
(Recei ed 17 May 2007; accep ed 25 Sep embe 2007)
Abs ac – A whole genome scan was ca ied ou o de ec quan i a i e ai loci (QTL) o
e ili y ai s in Finnish Ay shi e ca le. The mapping popula ion consis ed o 12 bulls and
493 sons. Es ima ed b eeding alues o days open, e ili y ea men s, ma e nal cal mo -
ali y and pa e nal non- e u n a e we e used as pheno ypic da a. In a g anddaugh e design,
171 ma ke s we e yped on all 29 bo ine au osomes. Associa ions be ween ma ke s and ai s
we e analysed by mul iple ma ke eg ession. Mul i- ai analyses we e ca ied ou wi h a a i-
ance componen based app oach o he ch omosomes and ai combina ions, which we e ob-
se ed signi ican in he eg ession me hod. Twen y- wo ch omosome-wise signi ican QTL
we e de ec ed. Se e al o he de ec ed QTL a eas we e o e lapping wi h milk p oduc ion QTL
p e iously iden i ied in he same popula ion. Mul i- ai QTL analyses we e ca ied ou o es
i hese effec s we e due o a pleio opic QTL affec ing e ili y and milk yield ai s o o linked
QTL causing he effec s. This dis inc ion could only be made wi h con idence on BTA1 whe e
aQTLaffec ing milk yield is linked o a pleio opic QTL affec ing days open and e ili y ea -
men s.
QTL / e ili y /dai y cow
1. INTRODUCTION
High e ili y in cows is economically impo an o dai y a me s. Low e -
ili y leads o highe eplacemen cos s, e e ina y cos s, labou cos s and cos s
due o educed milk p oduc ion. The p opo ion o e ili y ea men s ep e-
sen s 21% [36] o all he e e ina y ea men s in Finland. Also, 20% o he
in olun a y culling cases in Finland a e due o e ili y diso de s (Rau ala, pe -
sonal communica ion, 2004).
∗Co esponding au ho : nina.schulman@m . i
A icle published by EDP Sciences and a ailable a h p://www.gse-jou nal.o g
o h p://dx.doi.o g/10.1051/gse:2007044
196 N.F. Schulman e al.
Fe ili y ai s ha e a low he i abili y and a e o en difficul o measu e [31].
Gene ic p og ess by adi ional b eeding can he e o e be slow and he neg-
a i e co ela ions wi h p oduc ion ai s a e o special conce n [34]. Pösö and
Män ysaa i [34] ha e epo ed ha a gene ic imp o emen o 500 kg milk yield
would inc ease cases o o ula o y diso de s by 1.7%-uni s and days open by
4.2 days. These a e ai s o which ma ke -assis ed selec ion could inc ease
gene ic p og ess compa ed o adi ional b eeding schemes [25,38].
A emp s ha e been made o map loci affec ing e ili y. QTL ha e been de-
ec ed o o ula ion a e [4], winning [26], days open [39], non- e u n a e
and s illbi h [24], e ili y ea men s [15], and p egnancy a e [2]. In Finland,
mapping e ili y ai s is easible because he e is a good heal h da a eco d-
ing sys em wi h a da abase main ained by he Ag icul u al Da a P ocessing
Cen e L d.
Se e al s udies ha e ound un a ou able associa ions be ween milk p oduc-
ion ai s and e ili y ai s [23, 34, 37]. Cows wi h high milk yield eco ds
end o ha e poo e e ili y pe o mances han cows wi h mode a e o low milk
p oduc ion. Selec ion o high milk yield has led o longe in e als be ween
cal ing and he ollowing p egnancy and an inc ease in e ili y diso de s. In
o de o use ma ke in o ma ion o selec o be e e ili y wi hou comp o-
mising imp o emen in milk p oduc ion, mo e knowledge on he ch omoso-
mal egions affec ing bo h milk and e ili y ai s and he unde lying genes
is needed. Milk p oduc ion ai s and e ili y ai s a e co ela ed gene ically.
This gene ic co ela ion may be due o pleio opic QTL affec ing bo h ai s
simul aneously and/o o linked QTL each affec ing one ai . Fo effec i e
ma ke -assis ed selec ion, i is necessa y o dis inguish be ween a pleio opic
QTL and a linked QTL o a oid undesi able co ela ed esponses. The s an-
da d way o deciding how many QTL (ma ginal effec s) and hei in e ac ion
effec s should appea in he inal model elies on compa ing se e al models,
e.g. single- ai analysis wi h one o mul iple QTL models ollowed by mul i-
ai analysis wi h pleio opic o linked QTL models. The e a e wo limi a ions
o his app oach: i s , i allows he compa ison o nes ed models only; second,
i is no clea how o adjus he signi icance h eshold o each consecu i e
es [5]. Akaike in o ma ion c i e ion (AIC) [1] o Schwa z Bayesian in o ma-
ion c i e ion (BIC) [41] a e wo c i e ia ha do no equi e ha he compa ed
models be nes ed and hey ha e o en been employed o choose ma ke co a i-
a es o mul iple QTL mapping [16, 17] o o di ec ly es ima e QTL numbe
e.g. [3, 5, 7, 30, 42]. Piepho and Gauch [33] ha e in es iga ed model selec-
ion c i e ia ia simula ion. Thei esul s sugges ha ou o he conside ed
Fe ili y QTL in Finnish Ay shi e 197
c i e ia BIC has he bes p ope ies and can be used o he es ima ion o he
numbe o QTL wi h main effec s.
The objec i es o his s udy we e (i) o use he Finnish g anddaugh e de-
sign da a o map QTL o e ili y ai s (days open, e ili y ea men s, pa-
e nal non- e u n a e, and cal mo ali y in he Finnish Ay shi e popula ion);
(ii) o dis inguish be ween pleio opy and linked QTL when a egion is a -
ec ing mo e han one e ili y ai o a leas one e ili y ai and milk ai
iden i ied p e iously by Vii ala e al. [46].
2. MATERIAL AND METHODS
2.1. T ai s and popula ion
Days open (DO) is calcula ed as he numbe o days om cal ing o he ol-
lowing p egnancy. Fe ili y ea men s (FT) include in o ma ion abou e ili y
ea men s done by a e e ina ian wi hin 150 days a e cal ing and in o ma-
ion abou culling due o e ili y p oblems. Non- e u n a e (NRR) indica es
he abili y o a bull o make cows p egnan . I s e alua ion is based on he in-
semina ion o he bull’s semen o a andom se o cows and in his s udy, is
measu ed as he non- e u n a e wi hin 60 days om insemina ion wi h he
i s 500 insemina ions o a bull included in he da a. Cal mo ali y (CM) is
measu ed he e as a ai o he si e o he cow. I indica es he mo ali y a bi h
o he offsp ing o he daugh e s. The esponse a iables used in QTL mapping
we e b eeding alues ob ained om he Finnish Animal B eeding Associa ion
mainly om he e alua ion ca ied ou in au umn 2000. Fo NRR, he b eeding
alues om he e alua ion ca ied ou in sp ing 1996 we e used because he e
was no enough da a o he six oldes g andsi es in he yea 2000 e alua ion
o NRR.
B eeding alues o DO we e es ima ed using a epea abili y animal model
and o FT a epea abili y si e model. Reco ds om he i s h ee lac a ions
we e used. All bulls in he mapping popula ion had daugh e eco ds om all
h ee lac a ions. Fo CM a si e-g andsi e model was used. CM and FT we e
eco ded as bina y ai s. The he i abili y es ima es used o calcula ing he
b eeding alues we e 0.05 o DO, 0.01 o FT, 0.03 o CM, and 0.03 o NRR.
The milk yield ai s used o pleio opic and linked QTL analyses we e he
ollowing: milk yield 1s lac a ion (MY), p o ein yield 1s lac a ion (PY), a
yield 1s lac a ion (FY). Daugh e yield de ia ions (DYD) o igina ed om a
es day animal model.
A g anddaugh e design was used o QTL mapping. Twel e Finnish
Ay shi e hal -sib amilies we e geno yped. Only ele en o hem could be used
198 N.F. Schulman e al.
o he analysis o CM because he smalles amily did no ha e enough sons
wi h daugh e eco ds o his ai . The numbe o geno yped sons pe si e
anged om 21 o 82 wi h an a e age o 41 sons. The o al numbe o sons in
he popula ion was 493. The a e age numbe o daugh e eco ds pe bull was
496 o DO, 468 o FT, and 841 o CM.
2.2. Ma ke s and geno ypes
Ma ke s we e geno yped on all 29 bo ine au osomes. All a ailable sons
o he chosen bull si es we e yped. A o al o 169 mic osa elli es and wo
candida e gene SNP we e used. Ou o hese, 21 mic osa elli es we e new
compa ed o hose epo ed in p e ious s udies wi h he Finnish g anddaugh-
e design [40, 46]. Thus, ele en linkage maps we e ecalcula ed. The link-
age maps a e a ailable a h p://www.m . i/julkaisu /ca leq l. The numbe o
ma ke s pe ch omosome a ied om 2 o 14. The a e age spacing be ween
ma ke s was 19 cM. The o al leng h o he analysed genome was 2618 cM.
ANIMAP [12] o CRIMAP [13] we e used o cons uc he linkage maps. The
me hods o DNA ex ac ion, PCR eac ion p o ocols, and elec opho esis ha e
been desc ibed in p e ious s udies [10, 47].
2.3. S a is ical analysis
QTL analyses consis ed o he ollowing s eps: (1) a genome scan was
ca ied ou using mul iple linea eg ession o ou e ili y ela ed ai s;
(2) he signi ican QTL de ec ed om (1) and milk p oduc ion QTL de ec ed by
Vii ala e al. [46] ha o e lapped wi h he e ili y QTL we e eanalysed wi h
he a iance componen me hod using a single- ai model (STVC); (3) mul i-
ai pleio opic (MTP) and linked (MTL) QTL models we e analysed when
QTL o wo e ili y ai s o one e ili y ai and one milk yield ai [46]
we e de ec ed on he same ch omosome.
2.3.1. Reg ession me hod
Associa ions be ween ma ke s and ai s we e analysed using a mul iple
ma ke eg ession app oach [22]. The model used was he ollowing: yij =ai+
bixij +eij,whe ey
ij is he b eeding alue o bull j, who belongs o amily i,
aiis he polygenic effec o hal -sib amily i, biis he allele subs i u ion e -
ec o a QTL wi hin amily i, xij is he condi ional p obabili y o bull j
Fe ili y QTL in Finnish Ay shi e 199
o inhe i ing he i s haplo ype om si e i, and eij is he esidual. Signi i-
cance h esholds and P- alues o he F-s a is ic, we e ob ained by pe mu a-
ion, which was epea ed 10 000 imes o each ai and ch omosome sep-
a a ely [8]. Genome wise P- alues we e ob ained by Bon e oni co ec ion
Pgenome =1−(1 −Pch omosome)29, whe e 29 is he o al numbe o ch omo-
somes analysed.
A wo-QTL model was i ed in he eg ession analysis o hose ch omo-
somes ha had mo e han h ee in o ma i e ma ke s i one signi ican QTL
had been de ec ed and i he es ima ed QTL posi ions in he indi idual am-
ilies indica ed wo diffe en posi ions [44, 45]. Wi h he wo-QTL model, he
pe mu a ions we e done o es wo QTL s. no QTL. I his esul exceeded
he ch omosome-wise signi icance h eshold o 5%, he P- alue o wo QTL
s. one QTL was ob ained om a s anda d F able. The deg ees o eedom o
he F s a is ic we e he numbe o g andsi es as he nume a o and o al numbe
o offsp ing minus h ee imes he numbe o g andsi es as he denomina o .
2.3.2. Va iance componen me hod
Single- and mul i- ai QTL mapping based on he a iance componen
me hod was ca ied ou using he me hod desc ibed by Lund e al. [27]. The
ai s we e modelled using he ollowing linea mixed model wi h nqnumbe
o QTL:
y=µ+Zu +
nq

i=1
Wqi+e,
whe e yis a ec o o b eeding alues o DYD eco ded on ai s o each
geno yped son, µis a ec o o o e all ai means, Zand Wa e incidence
ma ices, uis a ec o o andom addi i e polygenic effec esul s om a com-
bined effec o backg ound genes, qiis a ec o o he effec s o he i h QTL,
and eis a ec o o andom esidual effec s. The andom a iables u,qiand e
a e assumed o be mul i a ia e no mally dis ibu ed and mu ually unco ela ed.
Fo de ails o he me hod see Lund e al. [27].
The a iance componen s we e es ima ed using he a e age in o ma ion
es ic ed maximum likelihood algo i hm [18] implemen ed in he so wa e
package DMU [29]. The es ic ed likelihood was maximised wi h espec
o he a iance componen s associa ed wi h he andom effec s in he model.
Maximising a sequence o es ic ed likelihoods o e a g id o speci ic posi-
ions yields a p o ile o he es ic ed likelihood o he QTL posi ion. The
in e al o QTL was es ima ed by one-LOD suppo [28].

200 N.F. Schulman e al.
2.3.2.1. IBD ma ices
The elemen s in he IBD ma ix a e a unc ion o he ma ke da a and he po-
si ion (p) o a pu a i e QTL on he ch omosome. He e we used he mos likely
ma ke linkage phase in he si e and compu ed he IBD ma ix using a ecu -
si e algo i hm [48]. The IBD ma ices we e compu ed o e e y 4 cM along
he ch omosomes and used in he subsequen a iance componen es ima ion
p ocedu e.
2.3.2.2. Tes s a is ics
Hypo hesis es s o he p esence o QTL we e based on he asymp o ic dis-
ibu ion o he likelihood a io es (LRT) s a is ic, LRT =–2ln(L educed −L ull),
whe e L educed and L ull we e he maximised likelihoods unde he educed
model and ull model, espec i ely. The educed model always excluded he
QTL effec o he ch omosome being analysed. The wo-QTL models we e
compa ed wi h one-QTL (null) models. Th esholds we e calcula ed using
he me hod p esen ed by Piepho [32].
2.3.2.3. Model selec ion be ween pleio opic and linked-QTL models
Since he pleio opic and he linked-QTL models a e no nes ed, he
Bayesian In o ma ion C i e ion (BIC) [20, 41] was used o e alua e which
model was a ou ed. The wo models in he p esen s udy en ail he
same numbe o pa ame e s and consequen ly he BIC simpli ies o
2logp(y|ˆ
θlinkageMlinkage)
p(y|ˆ
θpleio opyMpleio opy). I he wo models a e assumed equally likely ap i-
o i, he esul s using his c i e ia a e an app oxima ion o he pos e io p oba-
bili y o he pleio opic model ela i e o he pos e io p obabili y o he linked
QTL model (Bayes ac o ). We used he BIC calib a ion able by Ra e y [35]
o in e p e ing BIC es ima es. A BIC sco e o ⩾6 (model M1 s. M2) in-
dica ed s ong e idence o M1 o e M2. Ano he less o mal c i e ion used
o indica e which model is mo e likely, is he es ima ed co ela ion be ween
QTL effec s on he wo ai s ( Q12) om he pleio opic model. The a ionale
behind using Q12 is ha i he wo ai s a e unde he in luence o a biallelic
pleio opic QTL he ue alue o Q12 will be one.
Fe ili y QTL in Finnish Ay shi e 201
3. RESULTS
3.1. Days open
In he single- ai eg ession analysis, QTL o DO we e de ec ed on BTA1,
2, 5, 12, 20, 25, and 29 a ch omosome-wise 5% signi icance (Tab. I). The
single- ai model wi h a iance componen analysis (STVC) con i ms QTL
on BTA1 and 12 in he same egion o he ch omosomes (Tab. I). The wo-
QTL model wi h eg ession was i ed o BTA1 and 2. No suppo was ound
o his model o ei he ch omosome. In he analysis wi hin amilies he e
we e wo o i e amilies wi h ch omosome-wise signi ican F- alues pe ch o-
mosome. The posi ions o he highes F- alues on he ch omosomes we e no
consis en be ween amilies. The es ima ed allele subs i u ion effec s in hese
amilies anged om 0.7 o 1.5 s anda d de ia ions o EBV, which means 5.2 o
11.1 days.
3.2. Fe ili y ea men s
Wi h he eg ession analysis, QTL we e de ec ed on BTA1, 10, 15, 19, and
25 a ch omosome-wise 5% signi icance and on BTA5 and 14 a ch omosome-
wise 1% signi icance (Tab. I). The STVC analysis con i ms he QTL o FT
on BTA1. The wo-QTL model using eg ession analysis was signi ican o
BTA1, 5, and 14 (Tab. II). The s onges e idence o wo QTL was on BTA14.
The e we e one o ou amilies wi h ch omosome-wise signi ican F- alues in
he analysis wi hin amilies. The posi ions o he highes F- alues diffe ed be-
ween amilies. The allele subs i u ion effec s anged om 0.6 o 2.2 s anda d
de ia ions o EBV o 0.62% o 2.22% o ea men s.
On BTA1 and BTA25 he QTL posi ions in he ac oss amilies analysis o
DO and FT we e o e lapping. Fo bo h ch omosomes he QTL posi ions we e
a he end o he ch omosome, on BTA1 close o ma ke BMS4014 and on
BTA25 close o ma ke AF5 (Figs. 1 and 2).
3.3. Cal mo ali y
In he single ai eg ession analysis, QTL o CM we e de ec ed on BTA4,
6, 11, 15, 18, and 23 a 5% ch omosome-wise signi icance (Tab. I). The STVC
analyses did no con i m any o he QTL o CM, howe e , he QTL on BTA4
and 15 we e close o signi icance. The wo-QTL model using eg ession was
no suppo ed o any o he ch omosomes. In he analysis wi hin amilies
202 N.F. Schulman e al.
BTA1
0
0.5
1
1.5
2
2.5
3
3.5
1 163146617691106121136151
cM
F- alue
Figu e 1. P o iles o linea eg ession es s a is ics o BTA1 om single ai analysis
ac oss amilies. Quan i a i e ai loci we e de ec ed o days open and e ili y ea -
men s . The uppe ho izon al line indica es he ch omosome-wise 5% h eshold le el
o e ili y ea men s and he lowe dashed line he ch omosome-wise 5% h eshold
le el o days open.
BTA25
0
0.5
1
1.5
2
2.5
3
3.5
4
1 4 7 10131619222528313437404346495254
cM
F- alue
Figu e 2. P o iles o linea eg ession es s a is ics o BTA25 om single ai anal-
ysis ac oss amilies. Quan i a i e ai loci we e de ec ed o days open and e ili y
ea men s . The 5% h eshold le els o he ai s a e shown. The uppe ho izon al
line indica es he ch omosome-wise 5% h eshold le el o e ili y ea men s and he
lowe dashed line he ch omosome-wise 5% h eshold le el o days open.
Fe ili y QTL in Finnish Ay shi e 203
Table I. Quan i a i e ai loci o days open, e ili y ea men s, cal mo ali y and
non- e u n a e wi h eg ession and a iance componen me hods in Finnish Ay shi e
ca le.
T ai BTA1Reg ession me hod Va iance componen me hod
Pos.2(cM) F- alue Pos. (cM) LRT3
Days open 1 146 2.75∗∗ 144 11.29∗∗
2 2 2.86∗∗ 0.1 3.26
5 108 2.86∗∗ 107 4.29
12 47 2.34∗48 8.49∗
20 1 2.44∗25.80
25 47 2.93∗∗ 45 5.19
29 4 2.27∗45 4.90
Fe ili y 1 151 3.09∗148 9.75∗
ea men s
5 113 3.94∗∗ 84 3.83
10 145 2.99∗25.83
14 67 3.46∗∗ 50 1.37
15 1 3.30∗120 4.09
19 1 3.19∗11.78
25 54 3.60∗∗ –<1.0
Cal 4 17 2.36∗16.60
mo ali y 6 93 2.71∗85 3.9
11 29 2.09∗16 2.75
15 115 2.08∗120 6.33
18 1 2.24∗–<1.0
23 3 2.02∗12.05
Non- e u n 10 68 2.06∗144 3.54
a e 14 29 2.14∗30 2.85
1BTA =Bos au us ch omosome.
2Pos. =posi ion.
3LRT =likelihood a io es s a is ics.
∗P<0.05; ∗∗ P<0.01.
he e we e wo o ou amilies wi h ch omosome-wise signi ican F- alues
pe ch omosome. Fo BTA15, h ee amilies had hei highes F- alues close
o ma ke MGTG13B. Fo BTA18, wo amilies had hei highes F- alues a
BMS1355 and wo be ween ma ke s BMS1355 and BMS2213. On he o he
ch omosomes wi h signi ican QTL in he ac oss amilies analysis, he posi-
ions o he highes F- alues we e no consis en be ween amilies. The allele
subs i u ion effec s o he de ec ed QTL anged om 0.5 o 2.2 s anda d de i-
a ions o EBV, which is 0.45% o 2.0% o CM.
210 N.F. Schulman e al.
The posi ion o he QTL on BTA10 [24] was be ween ma ke s TGLA378 and
TGLA102, which is close o ou inding nea he ma ke ILSTS53.
When compa ing he e ili y QTL wi h he posi ions o he milk ai QTL
de ec ed in an ea lie s udy in he same amilies [46], se e al milk and e ili y
QTL we e ound on he same ch omosomes. Fo example, on BTA25, whe e
QTL we e de ec ed o DO and FT a he end o he ch omosome, QTL o
milk yield and p o ein yield we e also de ec ed. Fu he mo e, BTA1, 2, 5, and
12 ha bou QTL o milk and e ili y on app oxima ely he same ch omosome
segmen s acco ding o he eg ession based linkage analysis.
Mul i- ai QTL analyses we e ca ied ou on eigh ch omosomes, which
ha bou QTL o mo e han one e ili y ai o a leas one e ili y ela ed
ai and one milk p oduc ion ai iden i ied ea lie by Vii ala e al. [46] in he
same popula ion. We selec ed he ch omosomes based on he signi icance o
he QTL in he eg ession analysis. Though some o hese QTL we e no sig-
ni ican in he single- ai VC me hod, we did no pu his as a p econdi ion o
selec ing he ch omosome and ai combina ions. This was done as a mul i-
ai analysis o a pleio opic QTL because i has highe s a is ical powe o
de ec ion and a highe p ecision o he es ima ed map posi ion compa ed o
analysing he ai s indi idually [19,21,43]. Sø ensen e al. [43] obse ed ha
his is especially ue when a second co ela ed ai wi h highe he i abili y
(e.g. milk yield ai s in ou s udy) is used oge he wi h a low he i abili y
ai (e.g. e ili y ai s in ou s udy). Besides, a majo i y o he QTL iden i-
ied by eg ession, which did no exceed he signi icance h eshold in STVC
analysis, had sugges i e e idence o QTL seg ega ion in he same egion o
he ch omosome when analysed wi h STVC. The e o e, we kep a libe al en-
y le el o he QTL o be included in he mul i- ai analysis. Ou esul s
suppo he ea lie indings o Jiang and Zeng [19], Kno and Haley [21] and
Sø ensen e al. [43] ha show ha mul i- ai analyses ha e mo e powe in
de ec ing QTL compa ed o single- ai analyses.
Mul i- ai QTL analyses we e able o dis inguish pleio opic QTL om
linked QTL only on BTA1 and no on he o he ch omosomes. The esul s
also indica ed linked QTL on BTA5, 12, 14 and 15 o e ili y ela ed ai s
and milk p oduc ion ai s, bu i was no possible o p ecisely selec he
linked model o e he pleio opic model o ice e sa. The QTL in e als
(one-LOD suppo ) on a single ch omosome affec ing mo e han one ai
we e la ge and o e lapping. Also, he seg ega ing amilies had QTL peaks
sp ead o e a conside ably la ge egion o he ch omosome. The ma ke den-
si y used in he genome scan was spa se (a e age ma ke spacing 19 cM) and
inc easing ma ke densi y may help in educing he QTL in e al in linkage

Fe ili y QTL in Finnish Ay shi e 211
mapping especially o dis inguish pleio opic/linked QTL. The ai s show a i-
able amoun s o gene ic co ela ion. A signi ican QTL o a gi en ai migh
be non-signi ican o a highly co ela ed ai bu s ill ha e an effec on i [11].
This makes he sepa a ion be ween a QTL ha ing a pleio opic effec on wo
ai s and a QTL affec ing only one ai and showing an effec on he o he
ai due o a linked QTL, difficul .
5. CONCLUSIONS
Fou ai s ela ed o bo ine e ili y we e analysed in a QTL mapping s udy.
A o al o 22 ch omosome-wise signi ican QTL we e sugges ed in eg ession
analysis and h ee we e con i med wi h he single ai a iance componen
me hod. Only ew o he de ec ed QTL ha e been epo ed in ea lie s ud-
ies and many o he QTL o he p e ious s udies we e no suppo ed in he
p esen s udy. This could be due o a low powe o de ec ion ela ed o he low
he i abili y and difficul y o adequa ely measu e hese ai s. Some o he e -
ili y ai QTL a e closely linked o milk p oduc ion QTL o he QTL show
pleio opic effec s on milk p oduc ion and e ili y ai s. A dense ma ke map,
la ge popula ion and linkage disequilib ium based mapping may be needed o
dis inguish wo-linked QTL om a pleio opic QTL.
ACKNOWLEDGEMENTS
We would like o hank Anneli Vi a, Jonna Roi o and Tiina Jaakkola o he
labo a o y wo k, Jukka Pösö om he Finnish Animal B eeding Associa ion
o p o iding he EBV o he bulls, and he AI s a ions o semen samples.
The esea ch was pa ly unded by he Minis y o Ag icul u e and Fo es y
o Finland (Resea ch g an 4750/501/2004) and he Finnish Animal B eeding
Associa ion.
REFERENCES
[1] Akaike H., A new look a he s a is ical model iden i ica ion, IEEE T ans.
Au oma . Con . 19 (1974) 716–723.
[2] Ashwell M.S., Heyen D.W., Sons ega d T.S., an Tassell C.P., Da Y., VanRaden
P.M., Ron M., Welle J.I., Lewin H.A., De ec ion o quan i a i e ai loci affec -
ing milk p oduc ion, heal h, and ep oduc i e ai s in Hols ein ca le, J. Dai y
Sci. 87 (2004) 468–475.
212 N.F. Schulman e al.
[3] Ball R., Bayesian me hods o quan i a i e ai loci mapping based on model se-
lec ion: app oxima e analysis using he Bayesian in o ma ion c i e ion, Gene ics
159 (2001) 1351–1364.
[4] Bla man A.N., Ki kpa ick B.W., G ego y K.E., A sea ch o quan i a i e ai
loci o o ula ion a e in ca le, Anim. Gene . 27 (1996) 157–162.
[5] Bogdan M., Ghosh J.K., Doe ge R.W., Modi ying he Schwa z Bayesian in o -
ma ion c i e ion o loca e mul iple in e ac ing quan i a i e ai loci, Gene ics
167 (2004) 989–999.
[6] Boicha d D., G ohs C., Bou geois F., Ce quei a F., Fauge as R., Neau A.,
Rupp R., Amigues Y., Bosche M.Y., Le eziel H., De ec ion o genes in luencing
economic ai s in h ee F ench dai y ca le b eeds, Gene . Sel. E ol. 35 (2003)
77–101.
[7] B oman K.W., Speed T.P., A model selec ion app oach o he iden i ica ion
o quan i a i e ai loci in expe imen al c osses, J. R. S a . Soc. B 64 (2002)
641–656.
[8] Chu chill G.A., Doe ge R.W., Empi ical h eshold alues o quan i a i e ai
mapping, Gene ics 138 (1994) 963–971.
[9] de Koning D.J., Pong-Wong R., Va ona L., E ans G.J., Giuff a E., Sanchez A.,
Plas ow G., Nogue a J.L., Ande sson L., Haley C.S., Full pedig ee quan i a i e
ai locus analysis in comme cial pigs using a iance componen s, J. Anim. Sci.
81 (2003) 2155–2163.
[10] Elo K., Vilkki J., de Koning D.J., Velmala R.J., Mäki-Tanila A.V., A quan i a i e
ai locus o li e weigh maps o bo ine ch omosome 23, Mamm. Genome 10
(1999) 831–835.
[11] Gau ie M., Ba celona R.R., F i z S., G ohs C., D ue T., Boicha d D., Eggen A.,
Meuwissen T.H.E., Fine mapping and physical cha ac e iza ion o wo linked
quan i a i e ai loci affec ing milk a yield in dai y ca le on BTA26, Gene ics
172 (2006) 425–436.
[12] Geo ges M., Nielsen D., Mackinnon M., Mish a A., Okimo o R., Pasquino A.T.,
Sa gean L.S., So ensen A., S eele M.R., Zhao X., Mapping quan i a i e ai loci
con olling milk p oduc ion in dai y ca le exploi ing p ogeny es ing, Gene ics
139 (1995) 907–920.
[13] G een P., Falls K., C ooks S., Documen a ion o CRIMAP, e sion 2.4,
Washing on Uni e si y School o Medicine, S . Louis, MO, 1990.
[14] G isa B., Coppie e s W., Fa ni F., Ka im L., Fo d C., Be zi P., Cambisano
N., Mni M., Reid S., Simon P., Spelman R., Geo ges M., Snell R., Posi ional
candida e cloning o a QTL in dai y ca le: Iden i ica ion o a missense mu a ion
in he bo ine DGAT1 gene wi h majo effec on milk yield and composi ion,
Genome Res. 12 (2002) 222–231.
[15] Holmbe g M., Ande sson-Eklund L., Quan i a i e ai loci affec ing e ili y and
cal ing ai s in Swedish dai y ca le, J. Dai y Sci. 89 (2006) 3664–3671.
[16] Jansen R.C., In e al mapping o mul iple quan i a i e ai loci, Gene ics 135
(1993) 205–211.
[17] Jansen R.C., S am P., High esolu ion o quan i a i e ai s in o mul iple loci ia
in e al mapping, Gene ics 136 (1994) 1447–1455.
Fe ili y QTL in Finnish Ay shi e 213
[18] Jensen J., Man ysaa i E., Madsen P., Thompson R., Residual maximum likeli-
hood es ima ion o (co) a iance componen s in mul i a ia e mixed linea models
using a e age in o ma ion, J. Ind. Soc. Ag ic. S a . 49 (1997) 215–236.
[19] Jiang C., Zeng Z.B., Mul iple ai analysis o gene ic mapping o quan i a i e
ai loci, Gene ics 140 (1995) 1111–1127.
[20] Kass R.E., Ra e y A.E., Bayes ac o s, J. Am. S a . Assoc. 90 (1995) 773–795.
[21] Kno S.A., Haley C.S., Mul i ai leas squa es o quan i a i e ai loci de ec-
ion, Gene ics 156 (2000) 899–911.
[22] Kno S.A., Elsen J.M., Haley C.S., Me hods o mul iple-ma ke mapping o
quan i a i e ai loci in hal -sib popula ions, Theo . Appl. Gene . 93 (1996)
71–80.
[23] K agelund K., Hillel J., Kalay D., Gene ic and pheno ypic ela ionship be ween
ep oduc ion and milk p oduc ion, J. Dai y Sci. 62 (1979) 468–474.
[24] Kühn C., Bennewi z J., Reinsch N., Xu N., Thomsen H., Loo C., B ockmann
G.A., Schwe in M., Weimann C., Hiendlede S., E ha d G., Medjugo ac I.,
Fö s e M., B enig B., Reinha d F., Reen s R., Russ I., A e dunk G., Blümel J.,
Kalm E., Quan i a i e ai loci mapping unc ional ai s in he Ge man Hols ein
ca le popula ion, J. Dai y Sci. 86 (2003) 360–368.
[25] Lande R., Thompson R., Efficiency o ma ke -assis ed selec ion in he imp o e-
men o quan i a i e ai s, Gene ics 124 (1990) 743–756.
[26] Lien S., Ka lsen A., Kleme sdal G., Våge D.I., Olsake I., Klungland H., Aasland
M., He ings ad B., Ruane J., Gomez-Raya L., A p ima y sc een o he bo ine
genome o quan i a i e ai loci affec ing winning a e, Mamm. Genome 11
(2000) 877–882.
[27] Lund M.S., Sø ensen P., Guldb and sen B., So ensen D.A., Mul i ai ine map-
ping o quan i a i e ai loci using combined linkage disequilib ia and linkage
analysis, Gene ics 163 (2003) 405–410.
[28] Lynch M., Walsh J.B., Gene ics and analysis o quan i a i e ai s, 1s Edn.,
Sinaue Associa es, Sunde land, 1998.
[29] Madsen P., Sø ensen P., Su G., Damgaa d L.H., Thomsen H., Labou iau R.,
DMU – A package o analyzing mul i a ia e mixed models, 8 h Wo ld Cong ess
on Gene ics Applied o Li es ock P oduc ion, Augus 2006, Belo Ho izon e,
MG, B azil, Book o Abs ac s, pp. 247.
[30] Nakamichi R., Ukai Y., Kishino H., De ec ion o closely linked mul iple quan i-
a i e ai loci using gene ic algo i hm, Gene ics 158 (2001) 463–475.
[31] Philipsson J., Gene ic aspec s o emale e ili y in dai y ca le, Li es . P od. Sci.
8 (1981) 307–319.
[32] Piepho H.P., A quick me hod o compu ing app oxima e h eshold o quan i a-
i e ai loci de ec ion, Gene ics 157 (2001) 425–432.
[33] Piepho H.P., Gauch J . H.G., Ma ke pai selec ion o mapping quan i a i e ai
loci, Gene ics 157 (2001) 433–444.
[34] Pösö J., Män ysaa i E.A., Gene ic ela ionships be ween ep oduc i e diso de s,
ope a ional days open and milk yield, Li es . P od. Sci. 46 (1996) 41–48.
[35] Ra e y A.E., App oxima e Bayes ac o s and accoun ing o model unce ain y
in gene alized linea models, Biome ika 83 (1996) 251–266.
214 N.F. Schulman e al.
[36] Rau ala H., Te eys a kkailun ulokse 2000, Nau a 31 (2001) 22–23.
[37] Royal M.D., Flin A.P.F., Woolliams J.A., Gene ic and pheno ypic ela ionships
among endoc ine and adi ional e ili y ai s and p oduc ion ai s in Hols ein-
F esian dai y cows, J. Dai y Sci. 85 (2002) 958–967.
[38] Ruane J., Colleau J.J., Ma ke assis ed selec ion o a sex-limi ed cha ac e in a
nucleus b eeding popula ion, J. Dai y Sci. 79 (1996) 1666–1678.
[39] Sch oo en C., Bo enhuis H., Coppie e s W., A endonk J.A., Whole genome scan
o de ec quan i a i e ai loci o con o ma ion and unc ional ai s in dai y
ca le, J. Dai y Sci. 8 (2000) 795–806.
[40] Schulman N.F., Vii ala S.M., de Koning D.J., Vi a J., Mäki-Tanila A., Vilkki
J.H, Quan i a i e ai loci o heal h ai s in Finnish Ay shi e ca le, J. Dai y
Sci. 87 (2004) 443–449.
[41] Schwa z G., Es ima ing he dimension o he model, Ann. S a . 6 (1978)
461–464.
[42] Siegmund D., Model selec ion in i egula p oblems: applica ion and mapping
o QTLs, Biome ika 91 (2004) 785–800.
[43] Sø ensen P., Lund M.S., Guldb and sen B., Jensen J., So ensen D., A compa ison
o bi a ia e and uni a ia e QTL mapping in li es ock popula ions, Gene . Sel.
E ol. 35 (2003) 605–622.
[44] Spelman R.J., Coppie e s W., Ka im L., an A endonk J.A., Bo enhuis H.,
Quan i a i e ai loci analysis o i e milk p oduc ion ai s on ch omosome
six in he Du ch Hols ein-F iesian popula ion, Gene ics 144 (1996) 1799–1808.
[45] Velmala R.J., Vilkki J.H., Elo K.T., de Koning D.J., Mäki-Tanila A.V., A sea ch
o quan i a i e ai loci o milk p oduc ion ai s on ch omosome 6 in Finnish
Ay shi e ca le, Anim. Gene . 30 (1999) 136–143.
[46] Vii ala S.M., Schulman N.F., de Koning D.J., Elo K., Kinos R., Vi a A., Vi a J.,
Mäki-Tanila A., Vilkki J.H., Quan i a i e ai loci affec ing milk p oduc ion
ai s in Finnish Ay shi e dai y ca le, J. Dai y Sci. 86 (2003) 1828–1836.
[47] Vilkki J.H., de Koning D.J., Elo K., Velmala R., Mäki-Tanila A., Mul iple ma ke
mapping o quan i a i e ai loci o Finnish dai y ca le by eg ession, J. Dai y
Sci. 80 (1997) 198–204.
[48] Wang T., Fe nando R.L., an de Beek S., G ossman M., Co a iance be ween el-
a i es o a ma ked quan i a i e ai locus, Gene . Sel. E ol. 27 (1995) 251–274.
[49] Win e A., K ame W., We ne F.A.O., Kolle s S., Ka a S., Du s ewi z G.,
Bui kamp J., Womack J.E., Thalle G., F ies R., Associa ion o a lysine-
232/alanine polymo phism in a bo ine gene encoding acyl-coA; diacylglyce ol
acyl ans e ase (DGAT1) wi h a ia ion a a quan i a i e ai locus o milk a
con en , P oc. Na l. Acad. Sci. 99 (2002) 9300–9305.