Gene . Sel. E ol. 39 (2007) 529–543 A ailable online a :
c
INRA, EDP Sciences, 2007 www.gse-jou nal.o g
DOI: 10.1051/gse:2007019 O iginal a icle
Changes in he exp ession o gene ic
cha ac e is ics ac oss coho s in skele al
de o ma ions o a med salmonids
An i Ka∗, Ossi Rb, Tuija Pb
aMTT Ag i ood Resea ch Finland, Bio echnology and Food Resea ch, Biome ical Gene ics,
31600 Jokioinen, Finland
bFinnish Game and Fishe ies Resea ch Ins i u e, Te o Fishe ies Resea ch and Aquacul u e,
72210 Te o, Finland
(Recei ed 2 No embe 2006; accep ed 24 Ap il 2007)
Abs ac – Gene ic analysis o diso de incidence in a med animals is challenged by wo ac-
o s. Diso de s in diffe en coho s and en i onmen s could be caused by diffe en ac o s, lead-
ing o changes in he i abili y and o less han uni y gene ic co ela ions ac oss coho s. Mo e-
o e , due o compu a ional limi a ions, liabili y scale he i abili ies a e y low incidence may
diffe om hose es ima ed a highe incidence. We es ed whe he hese wo dilemmas occu in
skele al de o ma ions o a med salmonids using mul igene a ion da a om he Finnish ainbow
ou b eeding p og amme and p e ious salmonid s udies. The esul s showed ha he i abili y
was close o ze o in coho s in which managemen p ac ices main ained incidence a a low
le el. When he e was a managemen ailu e and incidence was unusually high, he i abili y was
ele a ed. This may be due o compu a ional limi a ions a e y low incidence and/o because
de o ma ions a e induced by diffe en ac o s in diffe en coho s. Mos gene ic co ela ions
be ween de o ma ions eco ded in diffe en gene a ions we e weakly o s ongly posi i e. How-
e e , also nega i e co ela ions be ween gene a ions we e p esen , showing ha high liabili y a
one ime can be gene ically connec ed o low liabili y a ano he ime. The esul s emphasise
ha gene ic a chi ec u e o bina y ai s can be in luenced by ai exp ession.
animal b eeding /animal heal h /de o ma ion /he i abili y /salmonids
1. INTRODUCTION
Imp o ed a med animal wel a e is a undamen al issue o p esen -day con-
sume s. To ensu e p o i able and e hical a m animal p oduc ion and domes-
ica ion, animal wel a e mus be add essed. De elopmen al diso de s, such
as abno mal de elopmen o skele al s uc u es, occu among bo h wild and
∗Co esponding au ho : An i.[email p o ec ed]
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:2007019
530 A. Kause e al.
a m animals, bu he phenomenon is easily exagge a ed among a med ani-
mals. This is due o a leas h ee easons. Fi s , animals wi h de o ma ions
do no su i e well unde wild condi ions [28]. Second, ce ain de o mi ies
a ise specially in cap i i y because o he a i icial ea ing and managemen
sys ems [12, 18, 21]. Thi d, selec i e b eeding o high p oduc ion may ha e
de imen al effec s on heal h ai s [26, 31]. Consequen ly, he occu ence o
de elopmen al diso de s in a m animals should be con olled by means o im-
p o ed managemen p ac ices, and/o by selec i e b eeding when ai s ha e
gene ic a ia ion.
Animal b eeding ope a es on es ima ed b eeding alues, in o de o selec
pa en s wi h he highes gene ic po en ial [14, 24]. He i abili ies and gene ic
co ela ions be ween ai s a e needed when es ima ing b eeding alues. Mod-
e n quan i a i e gene ic s udies on a m and wild animals analyse mul igene -
a ional da a using animal models [14, 24]. To ob ain es ima es o he gene ic
pa ame e s, a ai eco ded on animals o igina ing om mul iple gene a ions
and mul iple en i onmen s (e.g., a ms) is ypically ega ded as a single ai
o which pheno ypic and gene ic a iances a e es ima ed. This p ac ice has
also been adop ed o he analysis o diso de s and mo ali y. Howe e , his
p ac ice may be unsui able i de o ma ions and mo ali y occu ing in diffe en
coho s and en i onmen s a e in ac diffe en ai s. This may be e ealed by
weak posi i e o e en nega i e gene ic co ela ions, and by diffe en amoun s
o gene ic a ia ion o a ai eco ded in diffe en coho s o en i onmen s.
Fa med salmonids a e well sui ed o s udy whe he de elopmen al diso de s
eco ded in diffe en coho s and en i onmen s exhibi diffe en gene ic cha -
ac e is ics. Da a om heal h eco ding sys ems o ish b eeding p og ammes
show ha he incidence o de o ma ions changes d as ically om place o place
and om gene a ion o gene a ion [10, 19, 23]. In Finland, skele al de o ma-
ions commonly occu a low backg ound incidence (≤2%). The low inci-
dence is a esul o a well-es ablished managemen sys em in which all majo
s esso s a e unde con ol. De o ma ions occu a high incidence in a e occa-
sions. This esul s om a majo ailu e in he managemen sys em caused, e.g.,
by a sp ead o a disease, unusually wa m wea he condi ions du ing incuba-
ion and ea ly g ow h phases, o inapp op ia e eed o mula ion [3,21,32,33].
The ac o s causing de o ma ions do no need o be he same in all coho s,
and de o ma ions induced by diffe en s esso s need no sha e common ge-
ne ic o igin. Thus, he e may e en be gene ic ade-offs(i.e., nega i e gene ic
co ela ions) be ween esis ance mechanisms agains al e na i e s esso s [5].
He e we es ed whe he skele al de o ma ions in a med ainbow ou
Onco hynchus mykiss eco ded in diffe en p oduc ion en i onmen s and in
Gene ic a ia ion o de o ma ions 531
diffe en gene a ions display cons an he i abili y and posi i e gene ic co e-
la ions be ween each o he . We es ed whe he liabili y scale he i abili y in-
c eases along wi h inc easing incidence o wo easons. Fi s , al hough he -
i abili ies o bina y ai s (e.g.,0=absen and 1 =p esen ) calcula ed on
he unde lying no mally dis ibu ed liabili y scale [6, 8, 34] should in gene al
be independen o he incidence [29], liabili y he i abili y e iden ly becomes
ze o when incidence is ze o. Thus, i is possible ha a e y low incidence, li-
abili y he i abili y can be educed due o compu a ional limi a ions. Second, i
is possible ha a an unusually high de o ma ion incidence le el, induced by a
single majo s esso , and amilies may show consis en diffe ences in esponse
o he majo s esso , leading o mode a e gene ic a ia ion o de o ma ions.
In con as , a backg ound incidence when all majo s esso s a e con olled
o , de o ma ions may ep esen me ely de elopmen al noise and/o may be
induced by a combina ion o diffe en ac o s wi h e y low incidence, bo h
ac o s leading o low he i abili y.
We u he assessed whe he posi i e o nega i e gene ic co ela ions exis
be ween de o ma ions eco ded in diffe en coho s and en i onmen s. I ai s
eco ded in diffe en coho s a e uly he same ai , gene ic co ela ions be-
ween he coho s end owa ds uni y. I de o ma ions in diffe en coho s a e
induced by diffe en s esso s, i is possible ha nega i e gene ic co ela ions
appea .
To es hese hypo heses, a se o incidences and liabili y scale he i abili ies
we e ob ained om he Finnish ainbow ou b eeding p og amme, and om
he p e iously published salmonid s udies [10,23]. De o ma ion eco ds om
disc e e gene a ions and en i onmen s we e ea ed as sepa a e ai s, allow-
ing us o eg ess he incidences on he he i abili ies, and o es ima e gene ic
co ela ions be ween he ai s.
2. MATERIAL AND METHODS
2.1. Popula ion s uc u e
The da a on he incidence o skele al de o ma ions we e ob ained om he
Finnish ainbow ou b eeding p og amme. The b oods ock ish we e held a
he esh wa e b eeding s a ion loca ed in Te o, Cen al Finland. The li e-
cycle o he ish is ully con olled, ish a e indi idually agged, and hei pedi-
g ee is main ained o e mul iple gene a ions [19]. The da a included a o al o
41 286 indi iduals in i e gene a ions ha ched in 1993, 1996, 1997, 1999, and
2000 (Tab. I). The pedig ee o all ish was known back o gene a ion 1990, in
532 A. Kause e al.
Table I. Popula ion s uc u e o each gene a ion.
Gene a ion
1993 1996 1999 1997 2000
nsi es, ndams 55, 93 75, 150 47, 108 65, 79 98, 122
ndams pe si e: mean ( ange) 4.5 (1–9) 2.0 (1–4) 2.3 (1–4) 2.9 (1–5) 2.0 (1–5)
nsi es pe dam: mean ( ange) 2.7 (1–3) 1.0 (1–1) 1.0 (1–1) 2.4 (1–3) 1.6 (1–3)
n ull-sib amilies 247 150 108 191 200
n amily anks 247 150 116 227 200
nsea es s a ions - - 2 - 2
n=Sample size.
which indi iduals we e assumed o be un ela ed. Gene a ion 1990 ish we e
pa en s o gene a ion 1993, which was u he used o es ablish gene a ions
1996 and 1997, which sha e 23 dams and 6 si es bo n in 1993. Gene a ion
1996 ish we e used o p oduce gene a ion 1999, and gene a ion 1997 ish o
p oduce gene a ion 2000.
In each gene a ion, ish we e ma ed a he esh wa e b eeding s a ion du -
ing Ap il–June using ei he nes ed pa e nal (gene a ions 1996 and 1999) o
pa ial ac o ial hal / ull-sib designs ( he o he gene a ions) (Tab. I). Each ime,
ma ings we e comple ed du ing a one o h ee week pe iod. In each gene a ion,
47–98 si es and 79–150 dams we e ma ed. Full-sib egg ba ches we e incuba ed
sepa a ely, and a he eyed-egg s age, each ull-sib amily was anspo ed o a
sepa a e 150 L amily- ank held indoo s. Ye , a o al o 68 and 41 ull-sib am-
ilies we e alloca ed in o wo anks in gene a ions 1997 and 1999, espec i ely
(Tab. I). Eggs ha ched in July.
Du ing he win e a e ha ching, inge lings we e emo ed om he am-
ily anks and agged using Passi e In eg a ed T ansponde s (T o an, L d.,
Ge many), allowing indi idual iden i ica ion h oughou he s udy. An a -
e age o 52 andomly chosen ish ( ange 40–72) om each ull-sib amily
we e agged and held a he esh wa e s a ion o g ow-ou . A e agging,
he ish we e ea ed oge he in he same ou doo aceway o one addi-
ional g owing season, and eco ded o skele al de o ma ions in Ap il–June.
The ish we e held in ea h-bo om aceways unde non-comme cial densi ies
(1.5–3.0 indi iduals·m−3).
Fo gene a ions 1999 and 2000, an addi ional a e age o 34 ( ange 23–86)
andomly chosen ish om each ull-sib amily we e u he agged and spli
in o wo g oups. These g oups we e anspo ed o wo sea s a ions loca ed
in he Bal ic Sea (Tab. I). These ish we e ea ed in he same ne -pen o
one addi ional g owing season, and eco ded o de o ma ions om Oc obe
Gene ic a ia ion o de o ma ions 533
Table II. Incidence, he i abili y (h2) and i s lowe (CLlow) and highe (CLhigh) 95%
pos e io densi y in e al o se en skele al de o ma ion ai s eco ded du ing i e
gene a ions and wo en i onmen s.
Loca ion/T ai Sample Incidence
Gene a ion abb e ia ion size (%) h2CLlow CLhigh
F esh wa e s a ion
1993 G93 7910 2.00 0.06 0.027 0.129
1996 G96 5818 2.03 0.04 0.021 0.113
1999 G99 3309 23.8 0.26 0.111 0.383
1997 G97 8112 1.78 0.04 0.022 0.104
2000 G00 6044 5.92 0.14 0.075 0.220
Sea s a ion
1999 G99sea 2516 8.19 0.20 0.081 0.330
2000 G00sea 7577 4.57 0.08 0.035 0.139
o Ap il. A sea, ish we e held in ne cages unde comme cial managemen
condi ions.
Reco ding o all ish in one gene a ion and one ea ing loca ion las ed 1 o
2 weeks. A he ime o eco ding, all ish had g own o wo g owing seasons,
and he ish a he esh and sea wa e s a ions weighed an a e age (±SD) o
1060 ±260 g and 1050 ±277 g, espec i ely.
2.2. T ai de ini ion
De o ma ions we e isually eco ded based on ex e nal cha ac e is ics o
he li e ish. De o med he e e e s o ish wi h de o med skele al s uc u es o
he head, neck, back o ail. A ish wi h any o m o de o ma ion was sco ed as
one and a no mal ish as ze o. Because he eco ding is ex e nal, he a e age
incidences gi en a e unde es ima es o he ue de o mi y a es. This is because
less des uc i e de o ma ions (e.g., usion o wo- h ee e eb ae) may no be
iden i ied by he ex e nal sco ing.
A o al o se en sepa a e ai s we e de ined. Each gene a ion, and o gene -
a ions 1999 and 2000 also, he sea and esh wa e en i onmen s we e ea ed
as sepa a e ai s (Tab. II).

534 A. Kause e al.
2.3. Gene ic analysis
We es ima ed a o al o se en he i abili ies and incidence le els sepa a ely
o gene a ions 1993, 1996, 1997, and o he sea and esh wa e en i onmen s
o gene a ions 1999 and 2000 (Tab. II). To es ima e he i abili ies and gene ic
co ela ions on he unde lying no mally dis ibu ed liabili y scale, we used he
MGP-DMU so wa e’s animal h eshold model applying a Bayesian s a is ical
app oach [20]. The animal model used o de o ma ions eco ded a he esh
wa e s a ion was:
yi=µ+animi+εi,
and o de o ma ions a he sea s a ions was:
yij =µ+animi+STATj+εij,
whe e µis a popula ion mean, animiis he andom gene ic animal effec (i =
numbe o animals), and STATjis he ixed sea es s a ion effec (j = wo sea
s a ions), εis a andom e o e m, and yis an obse a ion o an indi idual.
We u he es ed a andom ull-sib effec wi hou he pedig ee in o ma ion in
he model. This effec would ha e accoun ed o he a iance due o common
incuba ion and ini ial ea ing o ull-sib amilies, and o pa s o dominance
a iance. Howe e , his effec was negligible and hus we excluded i om he
models.
The so wa e used Ma ko chain Mon e Ca lo (MCMC) me hodology o
gene a e ma ginal pos e io dis ibu ions o he i abili ies and gene ic co e-
la ions [20]. He i abili ies we e ob ained om uni a ia e analyses. We an a
chain o 1 million MCMC i e a ions sepa a ely o each ai . F om he chain,
we sa ed e e y 20 h es ima e, p oducing a o al o 50 000 he i abili y es ima es.
Gene ic co ela ions we e ob ained om bi a ia e analyses. Because MCMC
sampling is ime consuming, o gene ic co ela ions we educed he chain
leng h so ha we ob ained 10 000 sa ed es ima es o co ela ions. The esidual
co ela ion be ween ai s was se o ze o in each bi a ia e analysis because
a single ish appea ed only in one gene a ion and one en i onmen . Fla p i-
o s we e used because we expec ed o ind diffe ences in he i abili ies o he
diffe en ai s, and po en ially bo h posi i e and nega i e gene ic co ela ions
be ween gene a ions [20,27].
F om he ma ginal pos e io dis ibu ions, we calcula ed he i abili y as h2=
VG(VG+VR)−1and gene ic co ela ion as G=COVGxy (VGx VGy)−0.5,whe e
x and y a e wo ai s, VGis gene ic a iance and COVGis gene ic co a iance.
The es ima ed gene ic (co) a iances include addi i e gene ic (co) a iances
bu also po en ial pa s o dominance and epis asis (co) a iances. The eade
Gene ic a ia ion o de o ma ions 535
should no e ha when calcula ing a co ela ion be ween de o ma ions eco ded
in diffe en gene a ions and simul aneously in diffe en en i onmen s, he e -
ec s o gene a ion and en i onmen a e con ounded.
When he i abili y eaches ze o, o gene ic co ela ion eaches uni y o minus
one, he pos e io dis ibu ion becomes skewed. In hese cases, we calcula ed
mode a he han mean alue, o desc ibe he mos common he i abili y o
co ela ion alue. A Bayesian 95% equal- ailed pos e io densi y in e al o
gene ic pa ame e s was calcula ed by iden i ying he cu poin s lea ing 2.5%
o he es ima es a he lowe and uppe ails o he ma ginal pos e io dis ibu-
ion [7].
3. RESULTS
3.1. Rela ionship be ween incidence and he i abili y
In he Finnish da a, he incidence o skele al de o ma ions anged om
1.78% o 23.8% (Tab. II), he highes alues being obse ed in gene a ions
1999 and 2000. The Finnish da a showed ha he i abili ies we e signi ican ly
inc eased when incidence o de o ma ions inc eased (Fig. 1), as hypo hesised.
This was e idenced by he eg ession model (adjus ed R2=95.6%, n=7)
wi h s a is ically signi ican linea (3.0±0.46, =6.56, p=0.0028) and
quad a ic eg ession coefficien s (−8.0±1.76, =−4.54, p=0.0105). The
highes he i abili ies we e obse ed in gene a ions 1999 and 2000 in esh and
sea wa e s a ions. The eade should no e ha he e is a conside able amoun
o unce ain y in he poin es ima es, and he 95% pos e io densi y in e als
o diffe en he i abili ies a e in mos cases pa ly o e lapping (Tab. II).
Incidence in he p e iously published salmonid s udies anged om 2.2% o
21.5% [10,23]. When he se en Finnish he i abili y es ima es we e combined
wi h he i e liabili y scale he i abili y es ima es om hese p e ious s udies,
he pa e n ound in he Finnish da a emained (Fig. 1). In he eg ession model
o he combined da a (R2=79.2%, n=12), bo h he linea (5.9±1.08,
=5.44, p=0.0004) and quad a ic eg ession coefficien s (−18.9±4.36,
=−4.34, p=0.0019) we e signi ican . These esul s indica e ha a low
backg ound incidence, he i abili ies we e close o ze o. In con as , a he inci-
dence o 7.5% o mo e, he i abili ies anged om 0.195 o 0.50.
Wi hin he ange o da a poin s a ailable, he he i abili ies seemed o each
a pla eau a he highe incidence. This may be a ue pa e n, bu he in luence
o he limi ed numbe o he i abili y es ima es as well as he conside able e o
a ound he poin es ima es (e.g., Tab. II) canno be excluded as an explana ion.
536 A. Kause e al.
0 5 10 15 20 25
Incidence o de o ma ions (%)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
He i abili y
Figu e 1. Rela ionship be ween incidence and liabili y scale he i abili y o skele al
de o ma ions in salmonids. Closed ci cle () indica es an es ima e om he Finnish
ainbow ou b eeding p og amme (n=7); open ci cle () indica es an es ima e o
ainbow ou by [23] (n=1 pai o incidence and he i abili y es ima es); open box ()
indica es an es ima e o A lan ic salmon by [10] (n=4). The s a is ically signi ican
quad a ic eg ession line is d awn h ough all da a poin s.
Thus, we do no eel con iden o discuss whe he he ela ionship is linea o
non-linea , o wha is he de ailed shape o he non-linea unc ion.
3.2. Gene ic co ela ions ac oss gene a ions
Thi een ou o 19 ac oss-gene a ion co ela ions o de o ma ion incidence
we e posi i e and six we e nega i e (Tab. III). On he one hand, he a e age o
he nine gene ic co ela ions be ween gene a ions 1993, 1996, 1997 and 1999
was 0.36 ( ange: 0.01–0.84). Fo h ee o hese co ela ions, he 95% Bayesian
equal- ailed pos e io densi y in e al did no include ze o. On he o he hand,
he nega i e gene ic co ela ions we e obse ed only o he pai s o co ela-
ions including gene a ion 2000. The a e age o he en gene ic co ela ions o
gene a ion 2000 wi h all he o he gene a ions was −0.19 ( ange: −0.90–0.27),
and he 95% pos e io densi y in e al did no include ze o o h ee o he
nega i e co ela ions (Tab. III).
Gene ic a ia ion o de o ma ions 537
Table III. Gene ic co ela ionsa,bbe ween gene a ions and en i onmen s.
T ai cG93 G96 G97 G99 G99sea G00
G93 •
G96 0.28 •
G97 0.16 0.17 •
G99 0.34 0.01 0.18 •
G99sea 0.71* 0.84* 0.59* 0.68* •
G00 −0.13 0.27 0.17 −0.42* −0.82* •
G00sea −0.10 0.24 −0.26 −0.90* 0.09 0.49
a*=95% pos e io densi y in e al does no include ze o.
bAll co ela ions diffe signi ican ly om uni y.
cT ai s a e de ined in Table II.
3.3. Gene ic co ela ions be ween en i onmen s
In gene a ion 1999, he gene ic co ela ion be ween de o ma ions eco ded
a he esh wa e and sea wa e s a ions was s ongly posi i e (0.68), and he
95% pos e io densi y in e al did no include ze o (0.42–0.85) (Tab. III). Fo
gene a ion 2000, gene ic co ela ion be ween he en i onmen s was posi i e
(0.49) bu he 95% pos e io densi y in e al included ze o (−0.01–0.84).
4. DISCUSSION
4.1. Changes in he i abili y
The i s majo inding o his s udy is ha imp o ed managemen condi ions
dec easing incidence o de o ma ions in salmonids also dec eased he amoun
o gene ic a ia ion a ailable o selec ion. Gene ic a ia ion o de o ma ions
was low o non-exis en a low backg ound incidence, while gene ic a ia ion
was ele a ed a unusually high incidence. This allows gene ic esponses o se-
lec ion o occu mos effec i ely in ex eme en i onmen al condi ions when he
incidence o de o ma ions is high. When enhancing a m animal pe o mance,
imp o ed managemen p ac ices and selec i e b eeding a e ypically comple-
men a y. Howe e , his was no he case o skele al de o ma ions in salmonids.
Ou esul s show ha he e is a ade-offbe ween effo s pu on managemen
o educe diso de incidence and gene ic po en ial o selec i e b eeding.
Two no mu ually exclusi e ac o s may explain his obse a ion. Fi s , he
gene ic pa ame e s we e he e es ima ed on he unde lying liabili y scale, and
hus, he magni ude o gene ic a ia ion is expec ed o be independen o in-
cidence [29]. Howe e , his does no need o apply when incidence is close