Relevanceo fexterio r appraisal inpi gbreedin g

CENTRALE LANDBOUWCATALOQUS

0000 0394 7351 Promotor: dr. ir. R. D. Politiek emeritus hoogleraar ind e veeteeltwetenschappen

Co-promotor: dr. ir. E. Kanis universitair hoofddocent vakgroep Veefokkerij lJNo$W> I'd5?

EvertJ .va n Steenbergen

Relevanceo fexterio r appraisal inpi g breeding

Proefschrift terverkrijgin gva nd egraa dva n doctor ind elandbouw - enmilieuwetenschappen , opgeza gva nd erecto r magnificus, dr.H .C .va nde rPlas , inhe topenbaa rt everdedige n opwoensda g6 jun i199 0 desnamiddag st evie r uuri nd eaul a vand eLandbouwuniversitei tt eWageninge n

/S/&=£&/?

BIBLIOTHEEK LANDBOUWUNIVERSITEITD WAGENINGEN

Van Steenbergen, E. J., 1990. Relevance of exterior appraisal in pig breeding (De waarde van exterieur beoordelingen in de varkensfokkerij). Doctoral thesis, Department of Animal Breeding, Wageningen Agricultural University. P.O. Box 338,670 0 AH Wageningen,Th e Netherlands. /^°2^t i3Sf

STELLINGEN

1. Een lineair beoordelingssysteem voor exterieurkenmerken bij varkens biedt goede mogelijkheden tot selectie op beenwerk. Dit proefschrift 2. De extra informatieva nvisuee lbeoordeeld e lichaamsmaten enlichaams ­ bouw aan het eind van de toetsperiode is voor selectie op mesterij- eigenschappen te verwaarlozen. Dit proefschrift 3. Om de functionaliteit van het beenwerk van op groei en spekdikte geselecteerdevarken s oppei l tehoude ndien td ehelf tva nd e getoetste beren opbeenwer k afgekeurd teworden . Dit proefschrift 4. Het beoordelen van de grootte en de evenredigheid van de achter- pootklauwen van opfokgelten met het oog op verlenging van de levens­ duur, heeft geen zin. Dit proefschrift 5. Selectie tegen afwijkende beenstanden bij opfokgelten heeft een positief effect op de gemiddelde levensduur van de zeugenstapel. Dit proefschrift 6. Voor een effectief selectiebeleid ten aanzien van beenwerk is voor diverse beenstandenkenni sva nd epositi eva nhe t populatie-gemiddelde ten opzichte vanhe t optimum noodzakelijk. Dit proefschrift 7. De zin van selectie op exterieur in een zeugenlijn ter verhoging van de gemiddelde levensduur, zal met name afhangen van de beenwerk- reputatie vanbetreffend e lijn. 8. De opbrengst als gevolg van extra genetische vooruitgang in de varkensfokkerij door invoeringva nhe t' animal-model ', hang t afva nd e juistheid van zowel de dier- als demodelkeuze . 9. Voor een optimale valorisatie van het varkenskarkas verdient de benutting van de aanwezige variatie in vleeskwaliteit meer aandacht danhe t terugdringen van die variatie.

10. Voorkeursbeleidjegen see nbepaald ebevolkingsgroe pmaak tva nhe tanti ­ discriminatiebeleid een lachertje. 11. Dewetenschappelijk e onderbouwingva nd eessentiël evoorwaard e voor de mogelijkheid van het ontstaan van leven door evolutie, namelijk een ouderdom van de aarde van minstens 4.6 miljard jaar, is gebaseerd op drijfzand. Monthy White, A.J., 1985.Ho e oud is de aarde? Evangelical Press, Durham, Engeland. 12. Effectieve selectie opvleeskwalitei t isno g een taaiehap .

Proefschrift van Evert J. van Steenbergen Relevance of exterior appraisal in pig breeding Wageningen, 6 juni 1990. Aan Tineke Natasja, Michiel en Larissa Aan mijn ouders VOORWOORD

Het in dit proefschrift beschreven onderzoek is uitgevoerd bij de vakgroep Veefokkerij van de Landbouwuniversiteit te Wageningen. De uitvoeringva nhe tonderzoe kwer dmed emogelij kgemaak t doord e financiële bijdrage van het Produktschap voor Vee enVlee s (P.V.V.) te Rijswijk. De varkensfokkerij-instellingen hebben een sleutelrol gespeeld door zorg te dragenvoo r alle exterieurbeoordelingen. Dehul p vanHendrix ' Voeders bij het vinden van de varkensbedrijven voor het praktijkonderzoek was van onschatbare waarde, evenals hetjaarlij kverzamele n en op tape zetten van de zeuggegevens.Aa n de twee jaar durende medewerking van de 24varkens ­ houders en een tiental inspecteurs heb ik bijzonder goede herinneringen overgehouden. Zonder de bijdrage van alle hiervoor genoemde instellingen enpersone n zouhe t onderzoek nietplaat s hebbenkunne nvinden . Graag wil ik daarom een ieder bedanken die op enigerlei wijze een bijdrage heeft geleverdaa nd etotstandkomin gva ndi tproefschrift .Ee naanta lva nhe nwi l ikhie r met namenoemen . Mijnpromoto r prof.dr.ir.R.D . Politiekbe n ikzee r erkentelijkvoo r de geboden mogelijkheden, zijn geduld en stimulerende invloed bij dit onderzoek. Zijn vertrouwen in dit onderzoek en zijn inbreng in de discussies heb ik zeer gewaardeerd. De aanzet tot dit onderzoek is gegeven door mijn directe begeleider en co-promotor dr.ir. E. Kanis. Hij is steeds nauw betrokken geweest bij de opzet,uitvoerin ge nverslaggevin gva nhe thie rweergegeve nonderzoek .Zij n kritische instellingheef tmed ebijgedrage naa nhe tto tstan dkome nva ndi t proefschrift. Egbert,hartelij k bedankt. De discussies in de begeleidingscommissie waren vaak verhelderd en inspirerend.Wieb eKoop se nHei nva nde rSteen ,hartelij kbedank tvoo r alle waardevolle suggesties en commentaar. Alfred de Vries, hartelijk bedankt voor allenuttig e discussies engezellig e gesprekken gedurende ald e jaren diew e gezamelijk in'onze 'kame r op Zodiachebbe n doorgebracht. Verscheidene personenhebbe nee nwaardevoll ebijdrag e geleverdvoo r het verbeteren van de Engelse tekst: Mike Grossman, Karin Meyer, Gary Rogers enOlwe n Southwood. Totslo twi l ikhe tLEB-fond sbedanke nvoo r definanciël e ondersteuning bij de afronding vanhe t proefschrift. CONTENTS

1.Introductio n 1

2.Descriptio nan devaluatio no fa linea rscorin gsyste m 13 forexterio rtrait si npig s (LivestockProductio nScienc e2 3 (1989):163-181 )

3.Geneti cparameter so ffattenin gperformanc ean dexterio r 37 traitso fboar steste di ncentra lstation s (LivestockProductio nScienc e2 4(1990) :67-85 )

4.Exterio ri nsows :1 .Effec to fparit ynumbe ran d 59 associationwit hreason sfo rdisposa l (LivestockProductio nScience :submitted )

5.Exterio ri nsows :2 .Relationship swit hreproductiv e 79 performancean dlongevit y

(LivestockProductio nScience :submitted )

6.Genera ldiscussio n 99

Summary 109

Samenvatting 115 Chapter 1

INTRODUCTION

General.

Thecharacteristic spig shav etoda yar eth eresult so fevolutio nan dth e selection that has taken place since their domestication. Foundation of pig herdbooks and start of central performance tests at the beginning of thiscentury ,hav ebee nimportan tstimul i forbreeder s toa mor e systematic selection in their breeding stock. Selection was mainly based on visual appraisal of external characteristics and functions (exterior traits). Organized appraisal of living pigs canb e traced back to livestock fairs orshow sstartin g inth eearl y1800' si nEurop ean dlate r inNort hAmerica . The main objective in most of the shows was to emphasize the ideal type offered by leading breeders at that time. Little consideration was given either to pedigree or individual performance. Some descriptions of the ideal type are givenb y Anderson (1938): "The retention of good size with a higher degree of quality is the goal for which breeders are striving; trimness of the jowl, under line and hams is a definite indicator of quality; thehai r coat is also an index of quality; a strong constitution is shown by a deep,wid e chest,well-fille dbac k of the shoulder, a body of good depth, a capacious middle,an d a good appetite".However , despite severe limitations of the show ring in rating animals according to their breedingvalues ,unti l about 1950n o suitablesubstitut ewa s developed for the evaluation of the general merit of theanimals .

Exterior traits can be characterized by measurements (e.g. width, height, circumference),photograph y (Calabotta et al., 1982) and visual appraisal of external characteristics and functions. From a viewpoint of objectivity and interpretation of the data, measurements and photography are attractive, whereas low measuring costs and quickness of measuring favourvisua l appraisal asmean s for characterization of exterior traits. Characterization of exterior isno t a goal in itselfbu t rather serves as an indicator of growth performance, reproductivity and constitution. This indication might be of interest when economically important traits (breeding goal traits) can not be measured on living animals (e.g. slaughter andmea t quality)o rwhe n traitsca nno tb emeasure d ata nearl y age (e.g. reproduction, longevity).

Exterior traits in relation toreproductiv e and growth performance.

Mainexterna l characteristics forreproductio nan dmaterna l abilityar e number and quality of the teats. Relationships of body measurements (length, width and height) and weight at the age of about 230 days with reproduction are negligible (Drobig et al., 1983;Schlege l et al., 1983). Until about 1950 estimation of slaughter quality on living animals was based only on visual appraisal of external characteristics. Since then, severalmeasurin gtechnique s forapplicatio no nliv eanimal swer edevelope d andenable d objective andmor eaccurat eestimate so fbackfa t thickness and leanmea t content (e.g. Price et al., 1960;Molenaar , 1985).

Exterior traits and constitution.

The term constitution refers to the general physical structure and condition of a body. This comprises for example locomotory ability and susceptibility todisease s andparasites .Constitutio n isconsidere d tob e important because it is seen as an indicator of longevity. Brody (1945) stated: "Functioni sintimatel ycorrelate dwit hstructur e andform .Th ear t of animal judging is crude because only the grosser aspects of form and structure areavailabl e forexaminatio nan dcomparison .Th eprincipl e that function is an expression of structure is,however , unassailable". Large changes inhousin g systems, feedingan d floor types during the last three decadeswer eaccompanie db ya considerabl e increaseo flea nmea tdepositio n efficiency. In the same period many reports dealing with constitutional problems,mainl ydu et ofoo tan dle gdisorders ,wer epublishe d (e.g.Smith , 1966; Grondalen,1974a ;Web be tal. ,1983 ;Rothschil dan dChristian , 1988). Legweaknes s isa well-establishe d termt odescrib e thesedisorder s and is defined as a clinical syndrome characterized by anomalies of attitude and gait and difficulties to rise and mount (Goedegebuure et al., 1980a). It is concluded that the main motive for exterior appraisal is its presumed relationship with constitution,particularl y with legweakness . Economics and factors related to legweaknes s are therefore discussed. Economic lossdu et ole gweaknes sca naris e indifferen tway sa teac h levelo fth eproductio npyramid : through (1)reduce dgeneti c improvement inperformanc e traits at nucleus level due to involuntary culling, (2) prematuredisposa lo fbreedin gstoc ka tnucleu san dmultiplie rlevels ,(3 ) losso fsale sa tnucleu san dmultiplie rlevels ,an d (4)reduce dreproduc ­ tionan dproductio nefficienc ya tal llevels . On central test stations 15 to 50%o f testedboar s are rejected for breedingbecaus e ofle gweaknes s and/or insufficient exterior (Smithan d Smith, 1965;Teusche r et al., 1972;Bereskin , 1979;Web b et al.,1983 ; C.B.V.Annua lReport , 1985). Walters et al. (1982)reporte d that 10Zo f rearedgilt so fa pi gbreedin gcompan yar erejecte dfo rsal et ocommercia l multiplierherd sbecaus e of insufficient "legs/feet"an d "conformation". Onmultiplie rherd s 10t o20 Ïo fdispose dsow sar eculle dbecaus e ofle g weakness/lameness (Dagorn and Aumaitre, 1979; Arthur et al., 1983;T e Brake, 1986;Dijkhuize n et al., 1989). A reduction of the legweaknes s problemca nincreas eaverag eher dlif eo fsows .Th eeconomi cbenefi to fa n increaseo faverag eher d lifefro mfou rt ofiv elitter s iscalculate db y Dijkhuizene tal .(1986 )an dD eVrie s (1989),an daverage dDf1 .45. -pe r sowpe ryear .Dijkhuize ne tal .(1989 )calculate d theeconomi c losseso f premature disposal of sows. The highest losses are found in cases of 'lameness/legweakness' ,averagin gDfl .183. -pe rculle dsow . Next toth eeconomi c consequences,le gweaknes sha sa negativ e impact onanima lwelfar ean demphasize sth enegativ eimag eo fintensiv ehusbandr y systems.Reductio no fth ele gweaknes sproble m istherefor e animportan t item.

In literature many factors are studied in relation to legweakness . Thosefactor sca nb edivide dint oenvironmenta lan dinterna l(i.e .derive d fromo rpar to fth eanimal )factors .Th emai nfactor swil lb ereviewe dher e after. Environmental factors. In the literature several factors concerning housingsyste m(e.g .typ eo fconfinement ,typ eo ffloors )ar eindicate dt o have asignifican t impacto nle gweaknes s (Satheran dFredeen ,1982 ;Va n derWa le tal. ,198 4) .A direc trelationshi po fannua lcullin grat ewit h typeo ffloorin g during gestation,siz eo fth eher dan dlactatio nlengt h isshow nb yD'Allair e etal . (1989). Several studies dealt with the effect of food supplements (biotin, sodiumbicarbonat e (NaHC03)an dvitami nC )o nle gweaknes s (Pennye tal. , 1980; Nakano et al., 1983;Bryan t et al., 1985a,b;Va nde rWa l et al., 1986).Sodiu mbicarbonat ean dvitami nC ar efoun dt ohav ea positiv eeffec t on locomotory ability while biotin showed an effect on severity ofto e lesions. Afeedin gleve leffec ti sdescribe db yKornega ye tal .(1983 )reportin g that structural soundness scores are at a more desirable level in ad libitum-fedtha ni nrestricted-fe dboars .Thi si sno tsupporte db yresult s ofGrondale n(1974a) ,Nakan oe tal .(1979 )an dCalabott ae tal .(1982) . Internal factors. Osteochondrosis is considered to be the major componento fth ele gweaknes ssyndrom e(Goedegebuur ee tal. ,1980a ;Nakan o etal. , 1981).Grondale n(1974a,b )foun dth edegre eo fosteochondrosi so f ananima lt ob erelate dt ole gweakness .H eals osuggeste dver ybroa dhams , alon gback ,an da narro wlumba rregio nt ob epredisposin gfactor sfo rle g weakness. Poor associations between the state of osteochondrosis of fatteningpig san dth edegre eo fle gweaknes sar epresente db yFredee nan d Sather(1978) ,Goedegebuur ee tal . (1980b),Va nde rVal ke tal .(1982 )an d Nakanoe tal .(1983) ,whil eReilan de tal .(1978 )state dtha tn ocorrela ­ tion existed between presence of osteochondrosis and leg weakness. Goedegebuuree tal .(1988 )conclude dtha talthoug hosteochondrosi sma ypla y arole ,i ti sno tth eprimar ycaus eo ffront-le gweaknes si nDuro cswine . Hilley et al. (1981)state d thatvisua l scores for legweaknes s dono t relatet oth edegre eo fosteochondrosis . Vaughan (1971)an dGrondale n (1974a)indicate d thatfas tgrowin gpig s appearedt ob epredispose dt ostructura lsoundnes san dlocomotio nproblems . This is confirmed by Nakano et al. (1983) who found a significant correlationbetwee naverag edail ygai nan dlocomotor yabilit yscores .Thos e resultsd ono tagre ewit hgrowt han dle gproble mfigure sfro mfeedin gleve l experiments,whic hindicat eeithe rni lo ra positiv eeffec to fgrowt hrat e on locomotory ability. Conflicting results might partly be due to differencesi nweigh tan dag ea twhic hanimal swer escore dfo rlocomotor y ability. It is generally believed that a constant selection for rapid growthan dmuscl emas sresult s ina sever eunbalanc ebetwee nliv eweigh t and the immature skeletal frame (e.g.Goedegebuur e et al. 1980a). That unbalancemigh tdecreas elocomotor yability . Heredityo fexterio rtraits .

Hetzer and Miller (1972)reporte dheritabilltie s ofmeasure s on live animals(length ,widt han dheight )rangin gfro m0.3 6t o0.48 .Heritabilit y estimates of totalnumbe r of teats averaged about0.3 0 (e.g.Pumfre ye t al., 1980).Geneti cbackgroun do fa larg enumbe ro findividua lle gtrait s isestimate db y Smith (1966)an dWeb b etal . (1983).Heritabilitie swer e low,bu t indicated thatsom ele gtrait smigh thav ea geneti cpredisposi ­ tion. Genetic evaluations of combined scores of leg traits, generally reporteda sle gweaknes so rsoundnes sscores ,resulte di nlo wt omoderat e heritabilities (Bereskin, 1979; Webb et al., 1983; Lundeheim, 1987). Rothschildan dChristia n (1988)showe da direc trespons e tofiv egenera ­ tionso fdivergen tselectio no nfront-le gweaknes si nDuro cpigs .Landrac e pigs are generally described tohav e poorer constitution thanYorkshir e pigs(Smith ,1966 ;Goedegebuur ee tal ,1980a ,Web be tal. ,1983 ;Lundeheim , 1987). These breed differences indicate genetic variation. It can be concludedtha trespons eo nselectio nfo rexterio rtrait si npig si slikel y tooccur .

Visualscorin go fexterio rtraits .

Inliteratur eman yreport so nvisua lappraisa lo fexterio ri nswin edea l withscorin gt oa nideal ,whil emos ttrait sar ecombination so findividua l traits. Information on objectively and conclusively defined distinct exterior traits is lacking which also holds for relationships between exterior traits and measures of a good constitution, i.e. longevity. Because importance of external characteristics is mainly through their presumed relationship with constitution, attention for exterior traits shouldb e focused on traits forwhic h arelationshi p withconstitution , particularly with leg weakness, is indicated by literature and/or pig breeders. Many differentmethod s for evaluation of exterior traits are inuse . Thompson et al. (1981, 1983) mentioned several advantages of a linear scoring system over a descriptive coding system: (1)degre e rather than desirabilityi srecorded ,(2 )score scove rth ebiologica lrange ,(3 )trait s are scored individually rather than incombination , (4)a wid e rangeo f numerical scores canb e used which allows analyseswit h continuous scale, (5)heritabilitie s forlinearl yscore dtrait sar ecomparabl e too rslightl y larger than corresponding traits scored in relation to an ideal and (6) linear scoring permits interpretation of biological relationships among exterior traits. In dairy cattle linear scoring systems are practiced widely. No literature reports, however, are found of linear scoring in pigs.

Investigations.

This thesis focuses on the consequences and relevance of selection on exterior traits that are presumed tob e related to constitution. Results andconclusion sar ebase do ninvestigation so fexterio rtrait so fcentrall y tested boars and of gilts and sows on 24 commercial multiplier farms. Investigations were setu p to: evaluate theus e ofa linear scoring system forexterio r traits inpig s obtainheritabilit y estimates of exterior traits estimate genetic and phenotypic correlations between exterior traits and growth performance estimate the relationships of exterior traits with reproductive performance, reasons fordisposa l and longevity indicate the importance of exterior traits forpi g breeding.

Before investigations started, exterior traits tob e investigated were chosen indeliberatio nwit h inspectors ofth eDutc hpi gherdboo k and three pigbreedin g companies. The choice of the traits tob e evaluatedwa sbase d on the expected economic value and the likelihood to have a relationship with constitution. For the present investigations 21 exterior traits are defined and scored on a linear scale which ranged from 0 to 9 with 0.5 intervals. Description and evaluation of the linear scoring method is reported inChapte r 2.

Heritability and genetic variance of a trait are main factors which determineamoun to fselectio nrespons efo rtha ttrait .Dependin go ngeneti c (co)variances, selection on one group of traits might also change the phenotypic and genetic value of other traits. Possibilities of selection forexterio r traitsan d impacto fcullin go nexterio r traitsfo rperfor ­ mancetrait sar ereporte d inChapte r3 .

Severity or degree of legweaknes s isno t linearly related with its economic relevance.Th emai neconomi c losses only occur if theleve lo f leg weakness is reached which necessitates premature disposal of the animal. Longevity ina breedin g herd,whic h among other factors isth e resulto fal lknow nan dunknow nmanifestation s ofle gweakness ,migh tb e agoo dindicato ro fle gweaknes s(Calabott ae tal ., 1982) .I npi gbreeding , constitution is especially relevant in dam lines because these lines provide sowsfo rcommercia lmultiplie rherds .Longevit y ofthes esow si s economically important (e.g.Dijkhuize n et al., 1986). Relationships of exteriortraits ,especiall ywhe njudge da ta nearl yage ,wit hlongevit yar e therefore investigated on commercial multiplier herds. In Chapter 4, exterior scores inrelatio n toag eo fsows , andassociatio n ofexterio r traitswit hreason sfo rdisposal ,ar ereported . Relationships of exterior traits with reproductivity, as well as significance of exterior traits and quantification of their effect on longevityar edescribe di nChapte r5 .

Possibilities for incorporationo flinearl y scoredexterio r traitsi n pigbreedin gprogram san deffect so nperformanc etrait sca nb ededuce dfro m Chapters2 an d3 .Th erelativ eimportanc eo fexterio rtraits ,derive dfro m their relationships with leg weakness and longevity, is indicated in Chapters4 an d5 .Practica lapplication so fth eresult so fchapter s2 t o5 arediscusse d inChapte r6 .

REFERENCES

Anderson,A.L. ,1938 .Swin eenterprises .J.B . Lippincott Comp.Chicago/ - Philadelphia,45 8pp . Arthur,S.R. ,Kornegay ,E.T. ,Thomas ,H.R. ,Veit ,H.P. ,Notter ,D.R . and Barczewski, R.A., 1983. Restricted energy intake for gilts during growth.III .Characterizatio no ffee tan dlimb san dsoundnes sscore so f sowsdurin gthre eparities .J .Anim .Sei .56 :876-886 . Bereskin, B., 1979. Genetic aspects of feet and legs soundness in swine. J. Anim. Sei. 48: 1322-1328.1 Brody, S., 1945. Bioenergetics and growth. Reinhold Publ. Co.,Ne w York, 1023pp . Bryant, K.L., Kornegay, E.T., Knight,J.W. ,Veit , H.P. and Notter, D.R., 1985a.Supplementa lbioti nfo rswine .III .Influenc e of supplementation to Corn- and Wheat based diets on the incidence and severity of toe lesions,hai ran dski ncharacteristic s andstructura l soundnesso fsow s housed inconfinemen t during fourparities .J . Anim. Sei. 60:154-162 . Bryant, K.L., Kornegay, E.T., Knight, J.W., Webb jr., K.E. and Notter, D.R., 1985b. Supplemental biotin for swine. I. Influence on feedlot performance,plasm abioti nan d toelesion s indevelopin gpigs .J . Anim. Sei. 60:136-144 . Calabotta,D.F. , Kornegay, E.T.,Thomas ,H.R. , Knight,J.W. , Notter, D.R., Veit, H.P., and Wright, M.E., 1982. Restricted energy intake and elevatedcalciu man dphosphoru sintak efo rgilt sdurin ggrowth .II .Gai t characteristics analyzed from16m mmotio npictur ephotography .J . Anim. Sei. 55:1395-1404 . C.B.V., 1985.Annua l report of aboa r test station (Jaarverslag selectie mesterij Zuid-Nederland, 1985). Dagorn,J . andAumaitre ,A. , 1979.So w culling: Reasons for and effect on productivity. Livest. Prod. Sei. 6:167-177 . D'Allaire, S.,Morris ,R.S. ,Martin ,F.B. ,Robinson ,R.A . andLeman ,A.D. , 1989. Management and environmental factors associated with annual sow culling rate:A path analysis. Prev.Vet .Medic. , 7:255-265 . DeVries ,A.G. , 1989.A mode l toestimat e economicvalue s of traits inpi g breeding. Livest. Prod. Sei. 21:49-66 . Dijkhuizen,A.A. ,Morris ,R.S .an dMorrow ,M ., 1986 .Economi c optimization of culling strategies in swine breeding herds, using the "PorkCHOP computer program". Prev.Vet .Medic , 4:341-353 . Dijkhuizen, A.A., Krabbenborg, R.M.M, and Huirne, R.B.M., 1989. Sow replacement: a comparison of farmers' actual decisions and model recommendations. Livest. Prod. Sei. 23:207-218 . Drobig, M., Ritter, E. and Ritze, W., 1983. Untersuchungen über Zusam­ menhängezwische nKörpermaße nsowi eAnsatzleistun gun dde rWurfleistun g beiSchweine nunterschiedliche r genetischerKonstruktion .Arch .Tierz . 26:173-185 . Fredeen,H.T .an d Sather,A.P. , 1978.Join tdamag e inpig s rearedunde r confinement.Can .J .Anim .Sei .58 :759-773 . Goedegebuure,S.A. , Häni,H.J. ,Va nde rValk ,P.C .an dVa nde rWal ,P.G. , 1980a.I .A morphologica linvestigatio no fth estatu so fosteochondrosi s inrelatio nt obree dan dleve lo ffeeding .Vet .Q .2 :28-41 . Goedegebuure, S.A., Van der Wal, P.G. and Van der Valk, P.C., 1980b. Osteochondrosis in pigs. II.Th e relationship between the status of osteochondrosisan dth edegre eo fle gweakness .Proc .6t hIPV SCongress , Mexico,p.326 . Goedegebuure,S.A. ,Rothschild ,M.F. ,Christian ,L.L .an dRoss ,R.F. ,1988 . Severity of osteochondrosis in three genetic lines of Duroc swine divergently selected for front-leg weakness. Livest. Prod. Sei. 19: 487-498. Grendalen,T. ,1974a .Le gweaknes si npigs .I .Incidenc ean drelationshi p toskeleta l lesions,fee dlevel ,protei nan dminera l supply,exercis e andexterio rconformation .Act aVet .Scand .15 :555-573 . Grondalen,T. ,1974b .Osteochondrosi san darthrosi si npigs .VII .Relation ­ ship to joint shape and exterior conformation. Acta Vet. Scand. 15 (suppl.46) :1-32 . Hetzer,H.O .an dMiller ,R.H. ,1972 .Correlate dresponse so fvariou sbod y measurements inswin eselecte dfo rhig han dlo wfatness .J .Anim .Sei . 35:743-751 . Hilley, H.D., Hill,M.A . and Penny,R.H.C. , 1981.Skeleta l system. In: Lemon,E.D. ,Glock ,R.D. ,Mengeling ,W.L. ,Penny ,R.H.C. ,Scholl ,E .an d Straw,B . (Editors), Diseases of swine,Iow aStat eUniv . PressAmes , Iowa,5t hedition ,pp.170-181 . Kornegay,E.T. ,Veit ,H.P. ,Knight ,J.W. ,Notter ,D.R. ,Bartlett ,H.S .an d Calabotta,D.F. , 1983.Restricte denerg y intakean delevate d Caan dP intake forboar s during growth.II .Foo tan dle gmeasurement s andto e andsoundnes sscores .J .Anim .Sei .57 :1182-1199 . Lundeheim,N. ,1987 .Geneti canalysi so fosteochondrosi san dle gweaknes s in the Swedish Pig Progeny Testing Scheme. Acta Agric. Scand. 37: 159-173. Molenaar, B.A.J., 1985. The use of real time linear array ultrasound scanners for evaluation of live body composition. 36th Annual Meeting of the EAAP,Halkidiki , Greece,abstrac t Pla.l. Nakano, T., Aherne, F.X. and Thompson, J.R., 1979. Effects of feed restriction, sexe and diethylstilbestrol and the occurrence of joint lesionswit h somehistologica l andbiochemica l studieso nth e articular cartilage of growing-finishing swine. Can.J . Anim. Sei. 59:491-502 . Nakano, T., Aherne, F.X. and Thompson, J.R., 1981. Leg weakness and osteochondrosis inpigs .Pi gNew s and Info.2 : 29-34. Nakano, T., Aherne, F.X. and Thompson, J.R., 1983. Effect of dietary supplementation of vitamin C on pig performance and the incidence of osteochondrosis inelbo w and stifle joint inyoun g growing swine.Can . J. Anim. Sei. 63:421-428 . Penny, R.H.C., Cameron, R.D.A., Johnson, S., Smith, H.A., Cole, J.P.L., Kenyon, P.J., and Taylor,J. , 1980. Foot rot of pigs:Th e benefits of biotin supplementation of sows. Proc. 6th IPVS Congress, Copenhagen. p.332. Price,J.F. ,Pearson ,A.M. , Pfost,H.B . andDeans ,R.J. , 1960.Applicatio n ofultrasoni c reflection techniques inevaluatio n fatness and leanness inpigs .J . Anim. Sei. 19:381-387 . Pumfrey,R.A. , Johnson,R.K. , Cunningham, P.J. and Zimmerman, D.R., 1980. Inheritance of teat number and its relationship tomaterna l traits in swine. J. Anim. Sei. 50:1057-1060 . Reiland, S.,Ordell ,N. ,Lundeheim ,N . andOlsson ,S.E. , 1978.Heredit y of osteochondrosis, body constitution and leg weakness in the pig. A correlative investigation using progeny testing. Acta Radiol, suppl. 358: 123-137. Rothschild, M.F. and Christian, L.L., 1988.Geneti c control of front-leg weakness in Duroc swine. I. Direct response to five generations of divergent selection. Livest. Prod. Sei. 19:459-471 . Sather, A.P. and Fredeen, H.T., 1982. The effect of confinement housing upon the incidence of leg weakness in swine. Can. J. Anim. Sei. 62: 1119-1128. Schlegel, W., Wähner, M. and Wiese, H.P., 1983. Untersuchungen zur Abhängigkeit der Fruchtbarkeitleistungen bei Jungsauen von Ihren Körpermaßenbe i Zuchtbenutzungsbegin. Arch. Tierz. 26:341-349 .

10 Smith,C. ,1966 .A not eo nth eheritabilit yo fle gweaknes sscore si npigs . Anim.Prod .8 :345-348 . Smith,R.G.C .an dSmith ,C. ,1965 .Repor to nboa rperformanc e testinga t Stirling.Anim .Prod .7 :284 . TeBrake ,J.H.A. ,1986 .Cullin g of sowsan d theprofitabilit y ofpigle t production.Neth .J .Agric .Sei .34 :427-435 . Teuscher,T. ,Weniger ,J.H .an dSteinhauf ,D. ,1972 .Di eBeinschwäch ebei m Schwein.Züchterisch e undproduktionstechnisch e Bedeutung.Schwelnez . Schweinemast20 :292-296 . Thompson, J.R., Freeman,A.E. , Wilson, D.J., Chapman, CA. andBerger , P.J., 1981.Evaluatio no fa linea rtyp eprogra mi nHolsteins .J .Dair y Sei.64 :1610-1617 . Thompson,J.R. , Lee,K.L .an dFreeman ,A.E. ,1983 .Evaluatio no fa linea r typeappraisa lsyste mfo rHolstei ncattle .J .Dair ySei .66 :325-331 . Vande rValk ,P.C. ,Va nde rWal ,P.G .an dGoedegebuure ,S.A. , 1982.Th e valueo fclinica lsymptom so fle gweaknes si nrelatio nt oosteochondro ­ sisi nswine .Proc .7t hIPV SCongress ,Mexico ,p.161 . Vande rWal ,P.G. ,Koomans ,P. ,Va nde rValk ,P.C .an dGoedegebuure ,S.A. , 1984.Le gweaknes s inpigs :Clinica lchange sattributabl e toth etyp e offlooring .Tijdschr .Diergeneeskd .109 :1038-1043 . Vande rWal ,P.G ., Hemminga ,H ., Goedegebuure ,S.A .an dVa nde rValk ,P.C ., 1986.Th eeffec to freplacemen to f0.30 %sodiu mchlorid eb y0.43 %sodiu m bicarbonatei nration so ffattenin gpig so nle gweakness ,osteochondro ­ sisan dgrowth .Vet .Q .8 :136-144 . Vaughan,L.C. ,1971 .Le gweaknes si npigs .Vet .Rec .89 :81-85 . Walters, J.R., Gray, J. and Penny, R.H.C., 1982.Experience s of api g breeding company: 1. gilt selection rates. Proc. 7th IPVSCongress , Mexico,p.323 . Webb,A.J. ,Rüssel ,W.S .an dSales ,D.I. ,1983 .Genetic s ofle gweaknes s inperformance-teste dboars .Anim .Prod .36 :117-130 .

11 Chapter 2

DESCRIPTION AND EVALUATION OFA LINEAR SCORING SYSTEM FOR EXTERIOR TRAITS IN PIGS

E.J. van Steenbergen

Department ofAnima l Breeding, Wageningen Agricultural University, P.O. Box 338,670 0A HWageningen, The Netherlands.

Published inLivestoc k Production Science 23 (1989): 163-181. Reproduced by permission of Elsevier Science Publishers B.V., Amsterdam.

13 ABSTRACT

Inpi gbreeding ,muc h attention ispai d to theanimal' s exterior,whic h is thought to have a predictive value for its constitution. The term constitution refers to the general physical structure and condition of a body.A n important part of constitution failures consists of leg and claw disorders. A linear scoring system for exterior traitsha sbee n set up to study relationships of exterior traitswit h (re)production and longevity. A description of judged exterior traits is given and the linear scoring systemha sbee nevaluate d interm sof :difference sbetwee n inspectors,us e of 19 (0 to 9 with 0.5 increments) categories, accuracy of the scoring method and relationships between traits. For that purpose 40 boars have been judged twice by each of 10 inspectors. Differences in mean and standard deviation occurred between inspectors, with most inspectors showing a marked preference for whole number categories. On average, repeatability was 0.60 and reproducibility, indicating the correlation between repeated measurements done by more than one person, was 0.30. Principal component analysis showed that 20 exterior traits could be summarized by five interprétable factors that explained 57%o f the total variance. Those factors couldb e characterized as1 )sid evie wo f legs, 2) way ofwalkin g and lifetime expectation, 3)size ,4 )harmon y of the frame and 5) rear legs andclaws .

Inpresen t form, usefulness of application of described linear scoring system for exterior traits for selection purposes higly depends on the persons who judge the animals. When selecting and training appropriate people satisfactory repeatabilities canb e achieved formos t traits.

INTRODUCTION

Selectiono nexterio r traitsha s traditionallyplaye da n important role inpi gbreeding . Before introduction ofultrasoni c measurement ofbackfa t thickness, selection on exterior traits was used to improve slaughter characteristics andconstitution ,wher econstitutio nrefer s toth e general physical structure and condition of abody . Afterwards more attention has beenpai dt oexterio r traitst oimprov econstitutio ni nparticular .A majo r constitutional problem is legweakness ,whic h is thought tob e correlated

15 with the exterior of the animal. Leg weakness is a clinical syndrome characterized by anomalies of attitude and gait, and by difficulties to rise and mount (Goedegebuure et al., 1980). It is an evaluation of the pig's ability to move, usually based on visual impression (Grondalen, 1976). Indirect selection on exterior traits is used to reduce forced culling of sows,boar s and slaughter pigs for legweakness . Forced culling of sows can cause considerable economic losses. Kroes andVa n Male (1979) reported a Dfl. 143 (=US $ 70)highe r income per sow per year for farmswit h a low (31%)annua l replacement compared to farms with a high (55%) replacement. Reasons for culling in sows, reviewed by Kunz (1986)indicate d twoimportan tcause s i.e.reproductiv e failure (45%) andle gweaknes s (15%).I ncentra l teststations ,1 5t o50 %o fteste dboar s are rejected forbreedin gbecaus e of legweaknes s (Smith and Smith,1965 ; Teuschere tal. ,1972 ;Bereskin ,1979 ;C.B.V .Annua lreport ,1985) . Culling oftes tstatio nboar s forle gweaknes s canreduc e selection intensity and, thus, selection response forproductio n traits. Someestimate so frelationship sbetwee nle gweaknes san dexterio r traits are available from the literature (Webb et al., 1983;Lodd e et al., 1985; Lundeheim, 1987; Rothschild et al., 1988). Grandalen (1974) found the shapes of the lumbar region and of the hind quarters to significantly influence locomotory ability in slaughter pigs. He noticed a marked relationship of anarro w concave lumbar regionan d ofbroa dhin d quarters with poor locomotory ability. Although in literature leg weakness scores concern legs and gait of the pig, definition and scoring systems for leg weakness were not uniform. No clear conclusion can be drawn from the literature. Many different methods for evaluation of legweaknes s and exterior are in use. Thompson et al. (1981, 1983) mentioned several advantages of a linear scoring system over a descriptive coding system: (1)degre e rather than desirability isrecorded , (2)score s cover thebiologica l range, (3) traitsar escore dindividuall y rathertha ni ncombination , (4)a wid e range of numerical scores can be used which allows analyses with continuous scale, (5)heritabilitie s for linearly scored traits are comparable to or slightly larger than corresponding traits scored in relation to an ideal and (6)linea r scoring permits interpretation ofbiologica l relationships among exterior traits.

16 Inanima lbreeding ,rankin go fanimal si sbase do nbreedin gvalue sfo r traitswhic har eo feconomi c interest.Whe na larg enumbe ro ftrait sar e tob econsidered ,the ninformatio nha st ob ecombined .Informatio no ntw o or more traits can be combined by a selection index procedure when phenotypican dgeneti cparameter san d(economic )weighin gvalue so fthos e traitsar eknown .However ,calculatio no fweight s forexterio r traitsi s hard, ifno t impossible.Principa lcomponen tanalysi sallow sa larg ese t ofvariable s tob e represented ina muc h smallernumbe r ofindependent , hypothetical variables ("factors"). The technique has proved useful in studieso nth ephenotypi crelationshi pbetwee nmeasure so fsiz ean dshap e ofliv eanimal s (e.g.Carpente r etal. ,1978 ;Arnaso nan dThorsteinsson , 1982; Zarneckie tal. , 1987).

In the Netherlands, an experiment has been set up to evaluate the relevance of selection for exterior traits in pigs. In theexperiment , exterior traits of pigs were described and defined on a linear scale. Duringtw oyears ,severa lthousan dgilt san dsow shav ebee nevaluate dfo r exterior traits to investigate the relationship with leg weakness, longevityan dso wproductivity .I naddition ,tes tstatio nboar shav ebee n evaluated toestimat eheritabilitie s ofexterio r traitsan dcorrelation s betweenexterio rtrait san dproduction . Thegoa lo fth epresen tarticl ei st odescrib elinearl yscore dexterio r traitsi npig san dt oevaluat eth elinea rscorin gmethod .

DESCRIPTIONO FA LINEA RSCORIN GSYSTE M

Thechoic eo fth etrait st ob eevaluate dwa sdetermine db yth eexpecte d economicvalu ean dth elikelihoo dt ohav ea relationshi pwit hle gweakness . In total, 19 singular and two composite traits (nipple quality and productive lifetime expectation) were described (Table 1).A further illustrationo feigh ttrait si si nFigur e1 .Trait swer edefine do na scal e of 19categorie s (0-9wit h 0.5 intervals). Inlinea r scoring systems in dairy cattle there isn o uniformity about the scale and the number of categoriest ouse ,althoug hmos tWes tEuropea ncountrie sus ea scal efro m 1 to9 (DeGraa fe tal. , 1987). Iti simportan twhe nusin gth e"9-scale " thatal lvariatio n inth etrait sbetwee nanimal sfind sexpressio n inth e

17 scores.A wid erang eo fnumerica lscore si seasie rt ohandl et oobtai nthi s variation.D eGraa f(persona lcommunication ,1987 )suggeste d2 0categorie s tob eoptimal .

Table1 .Descriptio no f2 1exterio rtraits .

Score 4*1

1.Connectiono fshoulde rblad ewit hbod y loose solid Back and loin 2.Length short long 3.Width narrow broad 4.Regularity irregular regular Hams 5.Length short long 6.Width narrow broad Forelegs 7.Frontvie wleg s 0-shaped X-shaped 8.Sidevie wleg s sickledstraigh t buckled 9.Sidevie wpaster n steepangl e lowangl e Rear legs 10.Rearvie wleg s 0-shaped X-shaped 11.Sidevie wleg s straight bowle g 12.Sidevie whoc kjoin t steep sicklehocke d 13.Sidevie wpaster n steepangl e lowangl e 14.Quantitysubcut .liqui da thoc kjoin t much nil 15.Knottinesso flegs ;surfac eo ftubercle s > 12cm2 6cm2 0cm 2 Claws rear leg 16.Ratiosiz einne ran doute rcla w < h >1 17.Toesiz e small large 18.Gaitpattern ;movement s slow(stiff ) easy,quic k 19.Swingingo fbac k nil much 20.Nipples small large 21.Lifetimeexpectatio n short long

At the start of the research project, inspectors were instructed to assumea populatio nmea no f4.5 ,whil eemphasi sha sbee npu to nus eo fth e wholescale .I nth edescriptio no fth etrait sword ssuc ha s"bad" ,"best" , "satisfactory","normal" ,etc .hav eno tbee nuse dt oavoi dvalu ejudgemen t byth eclassifiers .Regularit yo fbac kan dloi nha dt ob eappraise db yth e transitionchange sfro mshoulde rt obac kan dbac kt oloin .Thu si twil lb e anevaluatio no fth eharmon yo fth ebod yframe .Sid evie wo fth erea rleg s

18 Trait fe\ 6 ^Â V./V

1CItlLÏL

10

Figure 1.Drawing s to illustrate 8 exterior traits. (6)Widt h of hams, (7)fron tvie w forelegs,(8 )sid evie wforelegs ,(9 )sid evie w pasternforelegs ,(10 )rea rvie wrea rlegs ,(11 )sid evie wrea r legs,(12 )sid evie whoc kjoint ,(13 )sid evie wpaster nrea rlegs .

19 hasbee nappraise dbase do n thepostur e determined at the stiflejoint . Teats were appraised by size and quality (e.g. presence of inverted nipples). Becauseexterio r isexpecte d tob erelate dt oso wlife-produc ­ tivity,lifetim eexpectatio nwa sdefine da sth eexpecte dnumbe ro flitter s to be produced in the future. This expectation was based only on the exterior of the sow. In present trial however, boars were scored for exterior.I norde r toge ta scor efo rlifetim eexpectation ,boar sha dt o bejudge da si fthe ywer egilts .

MATERIALAN DMETHO D

Thescorin gsyste mwa sapplie dexclusivel ydurin gth etw oyear so fth e project by inspectors of herdbook and breeding companies. To avoid systematicchange si nstandar ddeviatio nan dmea no finspector sdurin gtha t twoyea rperiod ,n oattemp twa smad et ocoac ho rstandardiz ethem . Atth een do fth etw oyea rperiod ,a tria lwa sse tu pt oevaluat eth e linearscorin gsystem .Eleve nDutc hLandrac e (DL)an d2 9Dutc hYorkshir e (DY)boar swer ejudge dtwic ewithi non eda yb yeac ho f1 0inspector sfro m theherdboo kan dbreedin gorganizations .Boar swer eshow nt oth einspector s oneb yon ea ta rando morder .T oreduc eth echanc eo nrecognitio no fboars , timebetwee nfirs tan dsecon djudgin gwa svarie dfro mboa rt oboa rfro m. 5 to4 hours .Liv eweigh to fboar s rangedfro m9 7t o11 1k gan dthei rage s variedfro m15 1t o17 1days . Thefollowin gcriteri awer euse dt oevaluat eth escorin gsystem : (a)Mea nan dstandar ddeviatio nfo reac htrait . (b)Us eo fcategories . (c)Repeatabilitie san dreproducibilities . (d)Relationship sbetwee ntraits .

(a)Standar ddeviatio npe r inspectorwa suse da sa measur efo rth eus e of the scale by the inspectors.Homogeneit y of inspector variances was testedpe rtrai twit hth eBartlet ttes t(Snedeco ran dCochran ,197 7p.296) .

(b)Th efrequenc ydistribution ,summe dove ral ltrait san dinspectors , wasuse dt oexamin eth eus eo fth ecategories .

20 (c) To estimate the repeatability and reproducibility, variance components were estimated using thefollowin g linearmodel :

R I B Y.ij k /• + i + h + \-.i + I*»« + * jki + e1Jk where, (1)

YiJk -exterio r score onanima lk ,bree d idon eb yinspecto rj fi - overall mean

Ri - fixed effecto fbree d i(i-1,2 )

Ij - fixed effecto finspecto rj (j-l,...,10)

Bfc.i -rando m effecto fboa rk withi nbree di

I*RtJ -fixe d effecto finteractio nbetwee n inspectorj an dbree di

I*Bjkl— rando meffec to finteractio nbetwee ninspecto rj an dboa rk withi n

breedi eiJk -rando m error

Repeatability (r^ isdefine d asth eexten t towhic h observations onth e same animal, judged by the same inspector, correspond with each other, i.e., an intra-judge correlation. It is influenced only by measurement errors.Th eexten tt owhic h scoreso nth esam eanimal ,judge db ydifferen t inspectors, correspond with each other isdefine d asth e reproducibility

(r2), i.e.,a ninter-judg e correlation. Differences between r1 andr 2ar e causedb yth einspector*boa r interaction. Informulae :

Repeatability (rx) = 1- a\ / a\ot

Reproducibility (r2)= 1 - (<7j»B+ a\) / a\ot where,

<7j»B = inspector*boar interactionvarianc e component a\ = errorvarianc e component

7?ot - °l +"f* B + °l

= boar variance component

In some publications with repeatability andreproducibilit y estimates (Hetzer et al., 1950;Evans , 1978;Janse n et al., 1985)th e effect of

21 inspectorswa sconsidere dt ob erandom .I nthos esituation sinspector swer e trainedt ojudg euniformly .I nth epresen tstudy ,however ,inspector swer e fromherdboo kan dthre ebreedin gcompanies ,wher eeac horganizatio nha dit s ownsyste mt ojudg ean dt oevaluat eth eanimal' sexterior .I nth etw oyear s that the inspectors used the linear scoring system, no meetings were organized tounifor m scoring over inspectors.Therefor e the differences betweeninspector swer epartl ydu et othei rbackgroun dan dthu sconsidere d tob efixed .T omak eresult smor ecomparabl ewit hliterature ,th eeffec t ofinspector swa sals oconsidere d tob erando man d 'variance'component s were estimated for inspectors. The inspector 'variance' component was includedi nth etota lvariance .I nformulae :

Repeatability (rj) -1 - a\ / (a^ + a\ +af. R)

Reproducibility (rj)- a\ / (a|ot+ a\fff« + R)where , a\ —inspecto r 'variance'componen t

crf.R -inspector*bree dinteractio n'variance 'componen t

Variance components have been estimated with the VARCOMP procedure (METH0D-TYPE1)o fSA S (SASInstitut eInc. ,1985) .

(d) The correlation matrix was used to examine the observation structure,t ose ewhethe rtrait scoul db eomitte do rgroupe dtogether .Whe n ahig h correlationbetwee n two traits exists,almos tn o new information willb eadde db yjudgin g thesecon dtrait .Omissio no fon eo fth etrait s maythe nb econsidered . Principal component analysis was used to reduce the large set of exterior traits toa smalle rnumbe ro findependen t factors, inwhic hth e exterior traits will be grouped. In principal component analysis three 'basic' steps are carried out: (1) the computation of the correlation matrix,(2 )th eextractio no fth einitia lfactors ,an d(3 )th erotatio nt o a terminal solution. The configuration of the factor structure is not unique; one factor solution can be transformed into another without violatingth ebasi cassumption so rth emathematica lpropertie so fa give n solution.B yfacto rrotatio nsimpl ean dinterprétabl efactor sar esearche d forb y measuring how close correlation coefficients between traits and

22 factor are to0 o r±1 .Th enumbe ro ffactor st oretai n ischose narbitrari ­ ly, but ismostl ybase d on theeigenvalue s ofth ecorrelatio nmatrix .Whe n eigenvalues of the correlation matrix are in descending order, the first eigenvalue represents thevarianc e explained by the first factor. In this study factors were extracted with principal component analysis by the FACTORprocedur eo fSA S (SASInstitut eInc. ,1985) . Orthogonalrotatio nwa s by theVARIMA X option.

RESULTSAN D DISCUSSION

Mean and standard deviation

Information about means, standard deviations (s.d.) and ranges is in Table 2. For most traits the range in inspector means was less than 2.5 points. For two traits (6an d 14),th eoveral lmea ndeviate dmor e thanon e point from the scalemea n (4.5). Thiswa spartl y due toth eunequa l number of boars per breed, because for those traits large breed differences existed. The overall s.d. averaged 1.15. Three of 21 traits showed an overall s.d. smaller than 0.9. The standard deviationwithi n inspectors was lower than the overall s.d. This was due to the large differences between inspector means.Postur e ofbot h fore- and rear legs (traits 7-13) showed little variation. Almost no variation was observed in nipple size and quality. The value of judging this trait is therefore nil, so it was excluded from further evaluationo fth escorin gmethod .A relatively large standard deviation was observed for knottiness, swinging of the back and lifetime expectation. A large standard deviation could be the result of a distribution with two peaks which could occurwhe n inspectors interpret a trait as twodifferen t traits.However ,n oevidenc e ofa distributio nwit h more than one peak could be found. Standard deviations per inspector (averaged over boars and traits) ranged from 0.62 to 1.16. For four inspectors a standard deviation lower than 0.9 was observed. From the test on homogeneity of variances it was concluded that differences between inspectors were significant (P<0.005) for all traits.Withi n some traits, inspectors differed mainly due to one or two outliers. However, the identity of the outliers (inspectors) was

23 not consistent over traits. For each trait a regression of standard deviation and variance on means was carried out. All regression coef­ ficientswer efoun dt ob enon-significan t (P> 0.10) ,s oi twa sconclude d that no consistent relationship between variation and mean existed. A transformationwil ltherefor eno tstabiliz eth evariance .

Table2 .Distributio ncharacteristic so flinea rscore dtraits .

Exter iortraits 1 MEAN Standarddeviatio n

overall I-range2 overall inspect3 I-range2 1. Shoulderblad e 4.30 (2.4-5.2) 1.30 0.98 (0.6-1.5) Back and loin 2. Length 5.33 (4.2-6.2) 1.07 0.91 (0.7-1.1) 3. Width 5.23 (4.2-6.2) 1.05 0.93 (0.5-1.1) 4. Regularity 5.17 (4.2-6.0) 1.24 1.08 (0.4-1.5) Hams 5. Length 5.01 (4.1-5.8) 0.96 0.75 (0.6-1.6) 6. Width 5.65 (4.5-6.4) 1.29 1.13 (0.6-1.1) Forelegs 1. Frontvie wleg s 4.56 (4.3-5.3) 0.77 0.62 (0.1-1.1) 8. Sidevie wleg s 4.60 (3.7-5.4) 1.11 0.95 (0.6-1.6) 9. Sidevie wpaster n 4.31 (2.8-5.0) 1.14 0.94 (0.5-1.3) Rear legs 10. Rearvie wleg s 4.67 (4.5-5.2) 0.65 0.53 (0.0-1.0) 11. Sidevie wleg s 4.67 (3.0-5.5) 0.98 0.70 (0.2-1.2) 12. Sidevie whoc kjoin t 4.58 (2.7-5.6) 1.05 0.73 (0.4-1.1) 13. Sidevie wpaster n 4.25 (2.0-4.9) 1.04 0.65 (0.3-1.0) 14. Liquida thoc kjoin t 5.64 (4.6-7.3) 1.38 1.01 (0.4-1.4) 15. Knottinesso fleg s 5.43 (4.5-7.1) 1.43 1.21 (0.6-1.5) Claws rear leg 16. Ratiosiz eo fclaw s 4.95 (4.0-6.4) 1.36 1.06 (0.7-1.6) 17. Toesiz e 5.00 (3.9-6.7) 1.27 0.85 (0.6-1.2)

18. Gait;movement s 5.24 (3.7-6.0) 1.37 1.15 (0.7-1.6) 19. Swingingo fbac k 4.51 (2.6-5.9) 1.63 1.21 (0.8-1.7) 20. Nipples 4.51 (4.5-4.5) 0.08 0.06 (0.0-0.1) 21. Lifetimeexpectatio n 4.22 (2.9-6.0) 1.94 1.61 (0.9-2.0)

1 :Trait sar edescribe di n Table 1. 2 : 1 -range= lowes tan dh i ghestinspecto rmean . 3 : i nspect- averag eo f1 0 within -inspector standard deviations.

24 Use of the categories

From the frequency distribution in Figure 2, it can be seen that the 'half' categories were used less thanwoul d be expected, except for 4.5. Therelativel yhig hnumbe ro fobservation s incategor y4. 5 canb e explained by the instruction to the inspectors to assume a population mean of 4.5. In Table 3 are the percentage observations in the 'whole' categories, in 'half' categories except 4.5 and in category 4.5. Differences between inspectors canb e noticed. Traits considering fore- and rear legs (traits 7-13) were scored frequently in category 4.5, which coincides with the relativelylo wstandar ddeviation .Scorin gi n'half 'categorie swa savoide d by a substantial number of inspectors, probably due to the coding of categories (0, .5,1,... )an d to the totalnumbe r of categories. Changing thecodin g ofcategorie s might reduce thepreferenc e forcertai ngroup s of figures. No conclusion canb e drawn on the optimal number of categories.

PERCENT»« 20

JKX1CX3 0 0.5 I 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 OTEC0RY

Fieure2 .Distributio no fobservation ssumme dove ral ltrait san dinspectors .

25 Table3 .Percentag eobservation si nth e'whole 'categories ,'half 'excludin g 4.5an dcategor y4.5 ,pe rtrai tan dinspector .

Category Category Trait 'Whole* 'Half' 4.5 Inspect. 'Whole' 'Half' 4.5 1 62 26 12 1 59 16 25 2 66 20 14 2 86 12 2 3 64 25 11 3 67 15 17 4 64 25 11 4 51 26 23 5 54 16 19 5 58 31 11 6 59 32 9 6 56 16 28 7 44 12 44 7 51 20 29 8 54 21 25 8 52 35 13 9 61 17 22 9 50 26 24 10 37 10 53 10 74 8 18 11 55 20 25 12 47 23 30 13 50 15 35 14 71 20 9 15 70 25 5 16 64 24 12 17 67 19 14 18 69 18 13 19 70 18 12 20 ------21 70 24 6

Repeatabilitiesan dreproducibilitie s

Analysiso fvarianc e showeddifference sbetwee nD Lan dD Yfo rsevera l traits (P<0.01). For most traits, boar*inspector interaction was sig­ nificant (P<0.001).I nTabl e4 varianc e analysiswit hdegree s offreedo m and expected mean squares is given, assuming inspectors to be random. Variance components, repeatability and reproducibility estimates of exteriortrait sar ei nTabl e5 .Th eboar*inspecto rinteractio nshowe dlarg e variance components for shoulder blade, regularity of back and loin, dryness of hock joint and lifetime expectation. A largeboar*inspecto r interaction variance component points to the possibility of different ranking ofboar sb y inspectors.Regula rcoachin gan ddiscussion sbetwee n themwil lprobabl yreduc etha tinteraction .Th eboa rvarianc ecomponen twa s largertha nth eerro rcomponen tonl yfo rwidt ho fth eham san dknottines s ofth erea rlegs .

26 Table 4. Analysis ofvarianc e andexpecte d mean squares.

Source d.f. Expected Mean Squares

Breed (R) 1 a\ +2crf. B+ 31.9crf.R+ 20a|+ Q(R )

Inspector (I) 9 a\ + 2af.B+ 48.1CT?.R+ 80cr f

Boar (B) 38 a\ + 2ff?.B+ 20

Inspector*breed (I*R) 9 a\ + 2af.B+ 31.9af. R

Inspector*boar (I*B) 342 a\ + 2CT|.B

Error (e) 400 a\

Repeatability (rx) estimates ranged from 0.33 (swinging ofth eback ) to

0.74 (widtho fth ehams )an daverage d 0.50.Reproducibilit y (r2)estimate s ranged from 0.12 (length of thehams ) to 0.62 (width of thehams )an d averaged 0.34.Lifetim eexpectation ,whic hi sdefine da sa composit e trait, can be judged with a relatively high repeatability. Repeatability and reproducibility estimates including aninspecto r 'variance' component (r^ and r2)average d 0.66an d0.25 , which is0.1 6highe r and0.0 9 lower than their respective estimates without the inspector 'variance' component.

Differences between r1 and r^ar edu e to differences in total variance

(o\ot). When considering inspectors random, total variance increases and thus r1 is higher than rx. Comparison of the formulae for r2 and r2 indicates that r2wil l always belowe r ora tmos t equal tor 2.Whe n using the scoring system in practice, uniformity of scoring will be pursued, which probably could induce a smaller inspector and inspector*boar interaction variance component. Intha t situation, repeatability (r^)an d reproducibility (r2) estimates would be between the estimates with and without aninspecto r 'variance' component. No repeatability estimates oflinea r scored exterior traits were found in the literature. De Roo (1983) investigated several exterior scoring systems that were used byherdboo k andbreedin g companies inTh eNether ­ lands. Hefoun d that r^average d 0.43an dr 2average d 0.33.Hetze r etal .

(1950)estimate d reproducibilities (r2)o feigh tbod y measurementso nliv e

27 Table5 .Varianc e components, repeatabilities and reproducibilities of linearscore dexterio rtraits .

Variancecomponents . insp* Exteriortraits . boar insp boar error r2 r2 1. Shoulderblad e .390 .746 .229 .400 .61 .38 .78 .22 Back and loin 2. Length .261 .292 .022 .432 .40 .37 .60 .24 3. Width .237 .178 .125 .314 .54 .35 .68 .24 4. Regularity .504 .311 .222 .513 .59 .41 .68 .32 Hams 5. Length .065 .305 .152 .325 .40 .12 .67 .07 6. Width .697 .324 .137 .298 .74 .62 .80 .46 Forelegs 7. Frontvie wleg s .094 .786 .066 .301 .35 .20 .49 .16 8. Sidevie wleg s .360 .266 .104 .473 .49 .38 .61 .30 9. Sidevie wpaster n .389 .373 .122 .429 .54 .41 .68 .29 Rear legs 10. Rearvie wleg s .100 .358 .071 .193 .47 .27 .54 .24 11. Sidevie wleg s .164 .425 .114 .298 .48 .28 .71 .16 12. Sidevie whoc kjoin t .119 .552 .150 .283 .49 .22 .75 .10 13. Sidevie wpaster n .127 .703 .058 .276 .40 .28 .76 .11 14. Liquida thoc kjoin t .336 .875 .283 .432 .59 .32 .78 .17 15. Knottinesso fleg s .708 .515 .194 .644 .58 .46 .70 .33 Claws rear leg 16. Ratiosiz eo fclaw s .485 .697 .177 .536 .55 .40 .72 .25 17. Toesiz e .233 .928 .113 .422 .45 .30 .75 .14

18. Gait;movement s .559 .520 .055 .720 .46 .42 .62 .29 19. Swingingo fbac k .426 1.217 .076 1.012 .33 .28 .63 .15 20. Nipples — ...... 21. Lifetimeexpectatio n 1.150 1.111 .404 1.164 .57 .42 .70 .30

pigs of approximately 100kg ;estimate s ranged from 0.56 to 0.77.Evan s (1978)investigate d theprecisio n ofbod y condition scoring on cowsan d ewes;r jwa s0.8 5 andr jwa s0.76 .Janse ne tal .(1985 )presente drepeat - abilities and reproducibilities ofbod ymeasurements ,fleshines s andfa t coveringo f15-mont hol dbulls ;fo rbod ymeasurement sr jrange dfro m 0.60 to0.7 8an dr £range dfro m0.5 0 to0.76 ,fo rfleshines sr ^wa s0.7 6an d r*2 0.14,an dfo rfa tcoverin gr jwa s0.8 0an d r\ 0.14. Althoughinspector spractice dth elinea rscorin gsyste monl ydurin gtw o years for approximately two days per month, repeatability and repro­ ducibility estimates for linear scored exterior traits were higher than

28 resultso fD eRo o (1983).Repeatabilit y andreproducibilit y estimates were muchlowe rtha nvalue sbase do nbod ymeasurement san dbod yconditio n scores (Hetzer et al.1950 ;Evan s 1978;Janse n et al. 1985).

Relationships between traits.

Pearson product-moment correlations for all exterior traits are in Table 6.Onl y nine of 210correlation s had an absolutevalu e greater than 0.5. The highest correlation (0.74)wa s between size and evenness of the claws. Apparently proportional and large size claws coincide.Th e highest correlationwit h lifetime expectationwa s forgai t (0.69)an d for swinging of thebac k (-.69). Easymovement s and a straightbac kwhil e walking were associatedwit h alon g lifetimeexpectation . Levelo fcorrelation s did not give aclea r indication for omitting one ormor e traitswithou t consider­ able loss of information.

Principal component analysis.

Table 7 presents the eigenvalues of the correlation matrix and the cumulative proportion of the total variance. Seven factors had an eigenvalue greater than unity. Together they explained 72% of the total variance. Thus a transformation of 20 traits to seven independent factors decreased the total variation by 28%.I n Table 8, the factor pattern of seven factors and thevariatio n explained by each factor after orthogonal transformation is presented. The first factor has high correlations with side view of the legs, about 0.82. The second factor can be seen as the way of walking and lifetime expectation. Size is mainly represented by factor three and harmony of the frame by four. Factor five provides some information about rear legs and claws. Factors 6an d 7ar emor e difficult to interpret. Therefore, it is concluded that 20 exterior traits can be transformed to five independent interprétable factors that explain 57%o f the total variance. Variance components of the exterior factors are estimated using model (1) and are presented with repeatabilities and reproducibilities inTabl e9 .Estimate srang efro m0.5 2 to0.6 7 forrepeat ­

ability (rt)an d from 0.41 to 0.53 for reproducibility (r2). Estimate s of ri> r2> r* and r2 f°r factor 1 and 5 (side view legs and rear legs/claws)

29 VO VO

I r-l

1 VO CTv in co o m vo en en CM O ON CM -1 O r-4 O o o 1-1 o 1-1 i-i O •H O o i-l

1 in m i-i av 1-1 o co rx m -tf m O 3 o r-l o o r-4 1-1 o r-l iH o iH r-l i-i CM CO

00 m en en CM co OO Cv CM i-i CM en r-l o o 1-4 o O CM CM o O •-I -a- en en

<* <* co 00 O ON VO CM en <* co i-i CM 1-1 O r-i en CM o i-i o O o o o en r o i-i en r-l o o» co in Oi I-l Ol m r-l i-i O o o i-i o o o r-l i-l O r-l i i g o en -d- i-i VO i-i r-l in -vt o o -i O o o 1-1 O r-l i-i O O CM

H o in rx 00 in r-l en m VO -l o o o O 1-1 CM in CM m vo

-1 o CM co vo erv 1-1 vo -1 o o o o r-l i-i en i-i vo

m in CM r-l m O O en 00 o o O r-l r-i 1-1 CM CM en o

ON CM m <* en en o r-l O o CM 1-1 o r-i

vo 00 in vo CM in CM 1-1 r-l o O CM O r-l I I 0\ M3 -cfr O r-l f f

CM m MD O

r-l

C e to O 01 co bil to 60 bf] C) to C) VU 1-1 C .rao 01 c ut r-l <1> m 01 CO O o O o tu o. o r-i r-l a. r-l r-l J2 P«. J3 P. CU X 10 Xi to co tu o 01 u 4-1 0) S S S S S S 4-1 to N « u •H •H m m o 01) (11 DJ 01 q U •ri •H •rl •rl •H •H ra c T) •rl —4 •rl •rl 01 •rl •rr-t o •u x: r-ral 4-> .c u > > > > > > •H 4-1 O to 00 r-l 4-1 1 rn of 4-1 3 ci: 4-1 G (Il u O CO CO 3 4-1 1-4 •rl 4-1 c o. CO O C T) 611 e Tl o T) T«) to T) T) T5 rr O t» 4-1 CO •rl •rl ex 14-1 n tl co •H CU 0) •rl 10 u •ri •ri em raCO 1- 1 •H •rl •rl C CO m O 1* •rl •r4 co £5 r-l S rr! r-l S bû Ut CO CO S 14 ai H ora C O •z. r-l <0 CU r-l a r-l -¥ CM en xt w m VO (1) r- CO CA 1H o rH CM en -•.T m vo N CO ON o r-l o H 14

30 arehighe r thanestimate s for itshighl ycorrelate d exterior traits,whil e values of the remaining three factors are comparable. Evaluation of exterior can be made less complex by analyzing a small number of factor scores compared to analyzing all traits separately. However, biological interpretation of relationships between those factors and (re)production might be more complex. The final choice of traits or factors to be considered in animalbreedin g depends onthei r relative economic value.

Table 7. Eigenvalues of the correlation matrix and the cumulative proportion of the totalvariance .

cumulative factor eigen­ proportion nr. value of variance

1 3.43 0.17 2 2.89 0.32 3 2.37 0.43 4 1.86 0.53 5 1.63 0.61 6 1.15 0.67 7 1.02 0.72 8 0.85 0.76 9 0.76 0.80 10 0.73 0.83 11 0.56 0.86 12 0.50 0.89 13 0.41 0.91 14 0.37 0.93 15 0.32 0.94 16 0.29 0.96 17 0.27 0.97 18 0.23 0.98 19 0.20 0.99 20 0.18 1.00

GENERALDISCUSSIO NAN D CONCLUSION

Scores of exterior traits were strongly influenced by inspector. Large differences between inspectors occurred formean s and standard deviations ofal ltraits .Fo rmos ttraits ,inspector*boa r interactionwa ssignificant . Interactionca nb ecause db ydifference s inscal ean dranking .Standardiza ­ tion of the scores, to remove scale differences, did not decrease the

31 Table8 •Facto r pattern of seven factors and thevariatio n explainedb y eacho fth efactor safte rorthogona ltransformation .

Factors

Exterio rtrait s 1 2 3 4 5 6 7

1. Shoulderblad e .04 .22 .09 .83 -.04 -.05 -.01 Back and loin 2. Length -.02 -.11 -.12 .26 .08 .78 -.05 3. Width .04 .03 .78 .23 -.04 -.15 .06 4. Regularity -.04 .17 .13 .84 .03 .14 .12 Hams 5. Length -.20 -.04 .57 .27 .02 .53 .14 6. Width .09 -.08 .88 -.02 -.04 -.09 -.10 Forelegs 7. Frontvie wleg s -.16 -.04 .03 .03 -.04 -.02 .83 8. Sidevie wleg s .28 -.12 -.18 -.20 -.06 .57 -.06 9. Sidevie wpaster n .66 .08 .27 -.15 .20 -.17 .22 Rear legs 10. Rearvie wleg s -.12 -.32 -.39 .20 .06 .01 .43 11. Sidevie wleg s .81 -.22 -.02 .02 -.15 .02 -.23 12. Sidevie whoc kjoin t .83 .04 -.04 .12 -.04 .11 -.14 13. Sidevie wpaster n .82 .17 .04 -.09 .19 .17 -.08 14. Liquida thoc kjoin t .09 .00 .07 .09 .85 .00 -.23 15. Knottinesso fleg s -.06 .06 -.16 -.01 .80 .01 .14 Claws rear leg 16. Ratiosiz eo fclaw s .20 .35 -.03 -.26 .48 .43 .08 17. Toesiz e .25 .37 .10 -.33 .47 .47 .17

18. Gait;movement s -.02 .81 -.02 .01 .13 .07 -.10 19. Swingingo fbac k -.04 -.79 .07 -.19 .03 .16 -.00 20 Nipples 21 Lifetimeexpectatio n -.01 .88 .07 .26 .07 -.08 -.02 Variationexplaine db y 2.74 2.62 2.06 2.00 1.97 1.77 1.19 each factor (eigenvalue)

interactioneffect .Evan s (1978)an dJanse ne tal .(1985 )foun danimal* - personinteraction sfo rbod ycondition ,fleshines san dfa tcoverin gscores . Inpractice ,exterio rscorin go fpig swil lgenerall yb eb ymor etha non e inspector.I nth esituatio no fn oinspecto rtraining ,reproducibility ,i n present study0.30 ,i sth ebes t indicator for theusefulnes s ofexterio r scoring. When no inspector differences would occur, then repeatability estimates,whic haverage d0.60 ,sho wa considerabl eincreas ei nusefulness . Carefultrainin go finspector san dperiodi cstandardizatio nexercise smigh t

32 Table 9.Varianc e components, repeatabilities (rx) and reproducibilities

(r2)o f five factors.

Variance components. insp* Factor boar insp boar error *i r2 *i r2

1. Sidevie w legs .156 .668 .076 .149 .61 .41 .86 .15 2. Walking/1if etim e .276 .349 .051 .305 .52 .44 .69 .28 3. Size .251 .192 .069 .156 .67 .53 .78 .35 4. Harmony frame .193 .570 .078 .194 .58 .42 .81 .18 5. Rear legs/claws .319 .329 .110 .209 .67 .50 .79 .32

Note: rj, r£ar e repeatability andreproducibilit y estimates including the inspectorvarianc e component.

decrease inspector differences and so increase theusefulnes s of a linear scoringsystem .Th e standarderro ro fa nanimal' s scoreca nals ob e reduced by scoring the animal a second time, preferable by a different person (Evans, 1978). In the Netherlands, a linear scoring system for exterior traits incow sha sbee nuse d for1 0years .Fo rtha tscorin g systemD eGraa f (personal communication, 1987) reported a correlation of 0.8 between exterior scoreso f the inspectorwh odetermine s the standard and scoresb y other inspectors. This too gives an indication that accuracy of exterior scores inpig s might be increased toa satisfactor y level. Principal component analysis has been shown to be useful in grouping correlated traits intoa smal lnumbe r ofinterprétabl e factors.Rankin g of animals on exterior might bemuc h easier when using those factors.

ACKNOWLEDGEMENT

Part of the costs of this studywer e covered by financial support from the Dutch Commodity Board for Livestock andMeat .

33 REFERENCES

Arnason, Th. and Thorsteinsson, S.S., 1982.Geneti c studies on carcass traitsi nicelan dtwi nra mlambs .II .Analysi so fprincipa lcomponent s andconstructio no fselectio nindices .Livest .Prod .Sei .8 :507-517 . Bereskin,B. ,1979 .Geneti c aspectso ffee tan dleg ssoundnes s inswine . J.Anim .Sei .48 :1322-1328 . Carpenter,Jr. ,J.A. ,Fitzhugh ,H.A. , Cartwright,T.C. ,Thomas ,R.C .an d Melton, A.A., 1978. Principal components for cow size and shape.J . Anim.Sei .46 :370-375 . C.B.V.,1985 .Annua l reporto fa boa r teststatio n (Jaarverslagselecti e mesterijZuid-Nederland ,1985) . De Graaf, F.M., Hewitt, D.E., Diers,H. , and Hamoen, A., 1987.Recom ­ mendationsfo rth eharmonizatio no ftyp eclassificatio no fth eEuropea n Black and White breed societies. Discussion report Meeting Conf. EuropeanBlac kan dWhit ebree dSoc. ,Brugge ,Belgium . De Roo, G., 1983. Selectie op exterieur bij varkens. Resultaten van vooronderzoeke nperspectieve nva nverder estudie .Interna lpublication , Dep.Anim .Breeding ,Agr .Univ .Wageningen . Evans, D.G., 1978. The interpretation and analysis of subjective body conditionscores .Anim .Prod .26 :119-125 . Goedegebuure,S.A. , Hani,H.J. ,Va nde rValk ,P.C .an dVa nde rWal ,P.G. , 1980.I .A morphologica linvestigatio no fth estatu so fosteochondrosi s inrelatio nt obree dan dleve lo ffeeding .Vet .Q .2 :28-41 . Grondalen,T. ,1974 .Le gweaknes s inpigs .I .Incidenc e andrelationshi p toskeleta l lesions,fee d level,protei nan dminera lsupply ,exercis e andexterio rconformation .Act aVet .Scand .15 :555-573 . Grendalen,T ., 1976 .Viewpoint so nth eporcin ele gweaknes ssyndrome .Proc . 3rdInter .Conf .Prod .Disease si nFar mAnimals ,Wageningen .p214-218 . Hetzer,H.O. ,Hankins ,O.G .an dZeller ,J.H. , 1950.Relationshi pbetwee n certainbod ymeasurement san dcarcas scharacteristic si nswine .J .Anim . Sei.9 :37-47 . Jansen,J. , BechAndersen ,B. ,Bergström ,P.L. ,Busk ,H. ,Lagerwey ,G.W . andOldenbroek ,J.K. , 1985.I nviv oestimatio no fbod ycompositio ni n youngbull sfo rslaughter .1 .Th erepeatabilit yan dreproducibilit yo f

34 a scoring system, anultrasoni c scanning technique andbod ymeasure ­ ments.Livest .Prod .Sei .12 :221-230 . Kroes, Y. and Van Male, J.P., 1979. Reproductive lifetime of sows in relationt oeconom yo fproduction .Livest .Prod .Sei .6 :179-183 . Kunz,H.J. ,1986 .Abgangsursache nbe iFerkel nun dSauen .Diss .Agr .Kiel . Lodde,K.H. , Dzapo,V . andWassmuth ,R ., 1985 .Analysi s of conformation objectionsi nselectio no fyoun gboars .II .Relation sbetwee nconforma ­ tionobjection s andperformanc e traitsan dth egeneti c dispositiono f conformationobjections .Züchtungskund e57 :47-57 . Lundeheim,N. ,1987 .Geneti canalysi so fosteochondrosi san dle gweaknes s inth eSwedis hPi gProgen yTestin gScheme .Act aAgric .Scand .37 :159-173 . Rothschild,M.F .an dChristian ,L.L. ,1988 .Geneti c controlo ffront-le g weakness in Duroc swine. I. Direct response to five generations of divergentselection .Livest .Prod .Sei .19 :459-471 . SAS Institute Inc., 1985.User' s Guide: Statistics,Versio n 5Edition . Cary,N.C. ,95 6pp . Smith,R.G.C .an dSmith ,C , 1965.Repor to nboa rperformanc e testinga t Stirling.Anim .Prod .7 :284 . Snedecor,G.W .an d Cochran,W.G. , 1977.Statistica l methods.Iow a state Universitypress ,Ames ,59 3pp . Teuscher,T. ,Weniger ,J.H .an dSteinhauf ,D. ,1972 .Di eBeinschwäch ebei m Schwein. Züchterische undproduktionstechnisch e Bedeutung.Schweinez . Schweinemast20 :292-296 . Thompson, J.R., Freeman, A.E., Wilson, D.J., Chapman, CA. andBerger , P.J., 1981.Evaluatio no fa linea rtyp eprogra mi nHolsteins .J .Dair y Sei.64 :1610-1617 . Thompson,J.R. , Lee,K.L .an dFreeman ,A.E. ,1983 .Evaluatio no fa linea r typeappraisa lsyste mfo rHolstei ncattle .J .Dair ySei .66 :325-331 . Webb,A.J. ,Rüssel ,W.S .an dSales ,D.I. ,1983 .Genetic so fle gweaknes s inperformance-teste dboars .Anim .Prod .36 :117-130 . Zarnecki,A. ,Ronningen ,K .an dStolzman ,M ., 1987 .Multivariat estatisti ­ cal methods for quantifying size and shape differences between Fl crosseso fdifferen tFriesia nstrains .J .Anim .Breedg .Genet .104 :28-34 .

35 Chapter 3

GENETIC PARAMETERS OF FATTENING PERFORMANCE AND EXTERIOR TRAITS OF BOARS TESTED IN CENTRAL STATIONS

E.J. van Steenbergen, E.Kani s andH.A.M ,va n der Steen

Department ofAnima l Breeding, WageningenAgricultura l University, P.O. Box 338,670 0A H Wageningen, The Netherlands.

Published inLivestoc k Production Science 24 (1990): 67-85. Reproduced by permission of Elsevier Science Publishers B.V., Amsterdam.

37 ABSTRACT

Geneticrelationship samon gfattenin gperformanc ean d2 0linearl y scored exterior traitswer e estimated from datao f279 2boar s of severalstrains . Boars were testedwit h ad libitum feeding from approximately 25 to 105 kg at central test stations of the Dutch herdbook and three breeding companies. (Co)variances offattenin gperformanc ean dexterio r traits were estimated simultaneously in a multi-variate model with the Restricted Maximum Likelihood method. Heritability estimates of performance traits ranged from 0.24 to 0.37 and of exterior traits from 0.01 to 0.38. High genetic correlations were found between daily gain and side view of the foreleg pastern, side view anddrynes s of thehoc kjoint .Slowe r growing animalshav e steeper foreleg pastern, less liquid at the hock joint and more sickle hocked rear legs. The genetic correlation between feed conversion and rearvie w of the rear legs indicates that boars with better feed conversion have more 0-shaped rear legs.Moderat eheritabilitie swer e found forfiv e linear combinations of exterior traits (factors). Index selectiononl y onfattenin gperformanc ewil l change the rearvie w of the rear legs to more 0-shape, change the side view of the hock joint tomor e steep, increase the quantity of liquid at thehoc k joint and slow down the boar's gait. A two-stage selection with selection in the first stage on gait (movements) and on growth performance in the second stage, reduces genetic progress of performance traits. When assuming a nihil economicvalu e forgai tpatter nan drejectin g 50%o fth eboar s inth e first stage, then economic merit is reducedb y 24X.

INTRODUCTION

In pig breeding, increasing attention has been paid to leg weakness during the last 20years .Althoug h nota nexac tdiagnosis ,le gweaknes s is awell-establishe d termt odescrib eth esyndrom eo flocomotor y disabilities commonly found inswin e (Grondalen, 1974). The legweaknes s status of live animals isbase d onvisua l appraisal of exterior traits, such as legs and claws, which are related to the pigs locomotory ability. The severity of thele gweaknes sproble mca nb e illustratedb y theincidenc e ofseriou s leg

39 weaknesswhic hi sgenerall yreporte di nth erang efro m1 5t o50 %(Smit han d Smith,1965 ;Teusche re tal. ,1972 ;Bereskin ,1979 ;C.B.V .Annua lReport , 1985). Culling on leg weakness can have a considerable effect on the geneticchang eo fperformanc etrait san dvic eversa ,dependin go ngeneti c (co)variances. Numerousestimate so fgeneti can dphenotypi cparameter so fperformanc e traits inpig shav ebee nreporte d inliteratur e (e.g.Sönnichse ne tal. , 1984a,b;Johansso ne tal. ,1985 ;Kenned ye tal. ,198 5an dMerks ,1987) .I n contrast,relativel y fewestimate so fsuc hparameter shav ebee nreporte d for exterior traits. Heritability estimates of leg weakness have been reportedi nth erang efro m0.1 0t o0.4 5 (Smith,1966 ;Reilan de tal. ,1978 ; Bereskin, 1979; Webb et al., 1983; Lundeheim, 1987; Rothschild and Christian,1988) . Results ofstudie sconcernin gth erelationshi pbetwee n leg weakness and production traits are contradictory. An antagonistic geneticcorrelatio nbetwee nle gweaknes san daverag edail ygai n(ADG )wa s foundb y Lundeheim (1987), while Bereskin (1979)an d Sather andFredee n (1982) found the association tob e beneficial.No n significant genetic correlationsbetwee nle gweaknes san dproductio ntrait swer efoun db ySmit h (1966), Webb et al. (1983) and Rothschild et al. (1988). The genetic correlation between leg weakness and backfat thickness is generally unfavourable (Bereskin, 1979; Webb et al., 1983; Lundeheim, 1987; Rothschild et al., 1988). Webb et al. (1983) did not find a genetic associationbetwee nfee dconversio nrati oan dle gweakness . Legweaknes s canb escore db yman ydifferen tmethods .Thompso ne tal . (1983) mentioned several advantages of a linear scoring system over a descriptive coding system. Relationships among linearly scored exterior traitsan d relationshipsbetwee nthos etrait san dperformanc e areeasie r tointerpret ,whil eheritabilit y estimatesar ecomparabl e too rslightl y larger than corresponding descriptively coded traits (Thompson et al., 1981; 1983). In dairy cattle, linear type appraisal has become common practice(Thompso ne tal .,1983 ;Swanso nan dBryan ,1984 ;Dier san dSwalve , 1987). Concerning pigs, no applications of a linear scoring system for exteriortrait swer efoun di nth eliterature . The present study consists ofa nanalysi s ofdat ao n linearly scored exteriortrait san dfattenin gperformanc etrait scollecte do nboar so fsi x strains.Th eobjectiv eo fthi sstud ywa st oestimat egeneti c(co)variance s

40 of exterior and fattening performance and to evaluate the impact of some selection strategies on exterior traits andperformance .

MATERIAL

Datawer eobtaine dfro mcentra ltes tstatio nboar so fth eDutc hherdboo k andthre ebreedin gcompanie sdurin g198 6an d1987 .Inspector so frespectiv e organizationsjudge dboar s atth een do fth etes ta tthei row ntes t station only. Each inspector judged abatc h of 10 to 30boar s per day. A batch is defined as boars of one strain judged within one day by one inspector. During the two years 20 to 30batche s were scored per inspector. Perfor­ mance testswer e approximately from 25 to10 5k gliv eweight .Th e exterior traits considered are described inth eAppendix .

Table 1.Numbe r ofboars ,sire san ddams ,averag e liveweigh t and standard deviation (betweenbrackets )a ten do fth e testan d thedistribu ­ tion of inspectors over datasets.

Strain1 boars sires dams weight (kg) Inspector

DL 638 82 461 104 (2.3) a,b + c,d ,e 2 DY 1043 124 831 104 (2.4) a,b + c,d,e2 A 207 36 131 98 (2.6) f.g Bl 366 59 197 109 (3.4) f.g B2 413 52 153 105 (5.3) h B3 367 54 160 118 (10.8) i

1 DL= Dutch Landrace; DY= DutchYorkshire ;A is a dam line and Bl to B3 are sire lines. Strain B3wa s group tested. 2 DL and DYwer e tested at two different test stations. Inspector a and b judged animals at station 1; c,d and ea t station2 .

Performance traits considered wereweigh t dividedb y age at end of the test (W/A), average daily gaino n test (ADG), average daily feed intake on test (DFI), average feed conversion (FC)expresse d as totalamoun to f feed consumed over total body weight gain on test and backfat thickness (UB)

41 ultrasonicallymeasure d atth een do ftest .Al lboar swer e feda d libitum. During the test boars were housed individually, except for one strain of whichanimal swer e group tested.A sa consequence ,F Can dDF Icoul dno tb e calculated in that strain. The structure of the data is shown inTabl e 1. About 8%o f thedam sha d offspring inth edat aoriginatin g from subsequent litters and different sires.Th e number ofboar swit h nil,on e and two or more full sibs is 1394,105 8 and 582,respectively .

METHODS

Each animalwa s judged for 21exterio r traits ona linea r scale of0 t o 9wit h0. 5 increments.A detaile ddescriptio no fmetho do fscorin g isgive n byVa n Steenbergen (1989). Intha t study, itwa s found thatnippl e quality scores showed almost no variation; therefore nipple quality was excluded from further analyses. Evaluation of the exterior by a large number of traits is difficult. Principal component analysis can be used to combine information of all traits into one or more 'factors', given the relationships between the traits. Van Steenbergen (1989) used principal component analysis to summarize these to f2 0exterio r traitst ofiv euncorrelate d factors.Base d on the factor loadings the factorswer e described as 1) 'side view legs', 2) 'movements and lifetime expectation', 3) 'size', 4) 'harmony of the frame'an d5 ) 'rearlegs' .Thos efiv efactor sexplaine d 57%o fth evarianc e of all exterior traits. Coefficients to calculate the five factor scores for the dataset described in the present paper, were adopted from Van Steenbergen (1989)an dar egive ni nth eAppendix .I nth epresen t studybot h exterior factors and exterior traits are analyzed. Variancecomponent sca nb eestimate db ysevera lmethod s (Kennedy, 1981). The choice of the method depends on structure of the dataset, computer facilities and desired properties of the estimates. Unbiasedness in variance component estimates, although a desirable property, is not of critical importance. Restricted Maximum Likelihood (REML), a biased procedure,ha sman y otherpropertie swhic h aredesirable ,no t theleas to f which is non-negativity of the variance component estimates (Kennedy, 1981). Increased computer capacity has enabled the use of REML on large datasetsbot hwit huni-variat e andmulti-variat emodels .Th ebenefi t of a

42 multi-variatemode l in comparisonwit h auni-variat e model relates to the (co)variance matrix which is defined tob e semi-positive definite, and to the gain in information through correlated traits. Because of the hierarchical structure of the data use of a multi-variate model implies exclusion of about 8% of the data. It was considered that the implied reduction in data did not offset thebenefit s of themulti-variat e model. (Co)variance component estimation was therefore carried out by a multi­ variate equal designREM Lprocedur ewit h theExpectatio nMaximizatio n (EM) algorithm, as described by Meyer (1985, 1986, 1987). This procedure involvesa transformatio nt ocanonica lscale ,changin ga n-variat e analysis ton uni-variat e analyses,whic hmake scalculation s feasible.However ,th e multi-variatemode lcoul dno thandl ea hierarchica lgeneti cstructur e (e.g. full sibs nested within half sibs groups). Therefore the dataset of 3034 animals was split into two sets (I and II). O f each full sib group two animalswer e randomly chosen and then randomly divided over the two sets. Animals without a full sibwer e assigned tobot h sets. Each set consisted of 2093 animals ofwhic h 1394wer e inbot h sets. In the model, that was used to estimate (co)variances, genetic groups (sires)wer e based on sires and paternal grandsires. The total number of random levels was 443.Th emodel :

Y 2 ijki- M + STi+ Gj:1+ bjCWiju.- W )+ b 2(Wljkl- W) + Sk:1+ eijkl included six strains (ST^ and 179 groups nested within strains (G,^) as fixedeffects ,liv eweigh t at endo f test (Wi,kl)linea r and quadratic with regression coefficients b1 andb 2 ascovariable , sireswithi n strain (Sk.j) as random factor and e^j^ the residual error associated with Y.,^, the record of the 1th progeny of the kth sire within strain i and judged in group j. The statistical model included weight at the end of the test to correct for the variation within strains, especially strain B3, as shown inTabl e 1. Van Steenbergen (1989) found differences between inspectors for both the mean and standard deviation of exterior traits. The model used to analyze thedat a only accounts fordifference s inmeans ,an d the inspector effect is also only relevant for exterior traits. Therefore a correction forheterogeneou s variancesbetwee n inspectorsan dstrain swa s carried out

43 for all exterior traits as proposed by Hill (1984). Within inspector and strain,exterio rscore swer etransforme dby : (x1Jkl- x ljk)/o- 1Jk +4.5 ,wher e xijkl*- stn e 1th observation of the jth inspector and the ith trait within th straink ,an d <71Jk and x^j. are the standard deviation andmea no f the j inspectoran dth ei thtrai twithi nstrai nk .B ythi stransformatio nexterio r differences between strains are also corrected for. Because of strain differences inperformanc e traits the straineffec twa s still included in themodel . (Co)variance componentestimatio nwa scarrie dou ti nbot hdatase t Ian d II and (co)variances were estimated simultaneously for exterior and performance traits.Startin gvalue s forvarianc e componentswer e obtained from uni-variate REML analyses using an EM-algorithm (Meyer, 1986, 1987) while covariances were set to zero. Iterationwa s stopped when the change ineac h of the (co)variance estimatesbetwee n successive roundswa s < 1%. (Co)variances of exterior factors were calculated from the (co)variances of exterior and performance traits,becaus e factors are linear functions of the exterior traits,b y the following formulae:

var(S?.1OiXi )= s;.!afvar(x i)+ 2E£^_i+1 o^covCx^Xj) (1)

cov(SJ.1 OL&, SJ.!ßjXj )= E^SJ. !a ißJcov(xi,xj) (2)

cov(2?=1 aLxit Xj)- 2C_Xa 1cov(xi,yj) (3) where, n = number of exterior traits (20) x.i>X j- exterior traits ian dj y_j - exterior orperformanc e traitj aL —loadin g of exterior trait io n a factor Bj — loading of exterior traitj ona n other factor.

Equation (1)denote sth efacto rvariances ,eqn .(2 )th ecovariance sbetwee n factors and eqn. (3)th e covariances between factors and exterior/perfor­ mance traits. Genetic parameters were based on (co)variance estimates averagedove rbot hsets .Heritabilitie s(h 2), genetican dphenotypi ccorrela ­ tionsbetwee n traitsx an dy (r andr respectively)wer eestimate das : 6xy pxy

44 2 h2 _ **8 . r _ -- . r _ -- 8 p al + a\ ' *y os * as *y o * o 00 x y *x vy Average standard errors of parameters for exterior and performance traitswer eapproximate db yth efollowin gformula :

J[(n/^n,^)) * (SE?+ SE?!) ] where,

ni snumbe ro fanimal si neac hdataset ;n tofci stota lnumbe ro fanimals ,an d SE?an d SE?jar e the squared standard errors of parameter estimates in datasetI an dII .Becaus eo fcommo nancestr ybetwee nth etw odatasets ,th e approximation isa nunde r estimation.Th e standard errors calculated in each dataset are the lower bound estimates.Averag e standard errors of parameterestimate sfo rexterio rfactor scoul dno tb eapproximated .

RESULTS

InTabl e 2,mean s and standard deviations forperformanc e traitsar e presented.Averag edail ygai nrange sfro m85 2t o99 2g .Th ehig hstandar d deviationfo rAD Go fstrai nB 3migh tb edu et oth ehig hstandar ddeviatio n ofweigh ta ten do ftes t (Table1 )becaus eo fth egrou ptesting .Strai nA hasth ethinnes tbackfa twhic hma yb erelate dt oth elowes tweigh ta ten d ofth etest . A significant effect of liveweigh t (linearan d quadratic)a ten do f testwa s foundfo rth eexterio r traitslengt han dwidt ho fbac kan dloi n andto esiz ean dfo rth eperformanc etrait sW/A ,AD Gan dFC . Estimates ofheritabilit y coefficients and of genetic andphenotypi c correlationso fperformanc etrait sar epresente di nTabl e3 .Heritabilit y estimatesrang efro m0.2 4 (DFI)t o0.3 7 (UB).Difference sbetwee ngeneti c andphenotypi ccorrelation sar esmall . Forexterio r factorsparamete r estimates arei nTabl e4 .Heritabilit y estimatesrang efro m0.1 3 ('movements/longevity')t o0.3 9 ('size'). 'Rear legs' isgeneticall ystrongl yrelate dt o 'movements/longevity'. Heritabilityestimate so fan dcorrelation sbetwee nexterio rtrait sar e inTabl e5 .Th eheritabilit yestimate srang efro m0.0 1 (knottinesso frea r legs)t o0.3 8 (widtho fbac kan dloin )an dal lstandar derror sar e< 0.07 .

45 Table2 .Mean s and standard deviations (in parentheses) of performance traits1b ystrain .

Strain W/A (g) ADG (g) DFI(kg ) FC UB(mm )

DL 634(39 ) 861(91 ) 2.36 (0.20) 2.75 (0.24) 12.6(1.7 )

DY 643 (43) 878(86 ) 2.25 (0.19) 2.57 (0.20) 11.1(1.4 )

A 657(47 ) 915(78 ) 2.17 (0.16) 2.38 (0.17) 10.4(1.3 )

BI 729(41 ) 992(65 ) 2.39 (0.18) 2.41 (0.15) 11.8(1.6 )

B2 679 (51) 925(79 ) 2.19 (0.20) 2.33 (0.14) 13.9(1.7 )

B32 651(62 ) 852(115 ) 14.0(1.8 )

W/A: weigh tdivide db yag ea tth een do fth etes t ADG: averag edail ygai n DFI: averag edail yfee dintak e FC :fee dconversio n(k gfee dpe rk gbod yweigh tgai na tth een do fth e test) UB :ultrasonicall ymeasure dbackfa tthicknes s Nofigure sfo rDF Ian dF Cbecaus eo fgrou pfeeding .

Postureo fth eforeleg s (traits7-8 )i sles sheritabl etha npostur eo fth e rearleg s(trait s10-12 ). Cla wcharacteristic shav ea lo wheritability .O n averagephenotypi ccorrelation sar emuc hsmalle rtha ngeneti ccorrelations . Standarderror so fphenotypi ccorrelation sar ealway s< 0.02 .Clos egeneti c correlations arefoun dbetwee nexterio r traitsreferrin g tosiz e (traits 2,3,5,6). Length and width are negatively correlated. Connection of shoulderblad ewit hth ebod yi shighl ycorrelate dwit hregularit yo fbac k

and loins (rg-=0.61 ±0.12; r=0.58) . Body width shows negative genetic correlationswit hsid evie wforeleg san dpositiv ewit hsid evie wforele g pastern;broade r animalshav e lessbuckle d forelegs andweake rpastern . Weak pastern forelegs are genetically related to X-shaped rear legs

(rg--0.50±0.15 ). Bowrea rleg sgeneticall ycoincid ewit hsickl ehocke dleg s

(rg=0.60±0.12) .Littl eliqui da tth ehoc kjoin ti sgeneticall ycorrelate d withwea kpaster no frea rleg s (r=0.5 3±0.15 )an da smoot hgai t (r=0.6 5 ±0.17).A geneti cassociatio ni sfoun dbetwee nlarg etoe so frea rleg san d

shortham s(r g=-0.47±0.19) ,les sbuckle dforeleg s(r g=-0.43±0.24 )an dwea k

46 Table3 .Heritabilitie s (onth ediagonal )an dgeneti c(belo wth ediagonal ) andphenotypi c (above thediagonal )correlation s ofperformanc e traitswit hstandar derro ri nparentheses .

W/A (g) ADG (g) DFI(kg ) FC UB (mm)

W/A 0.36 (0.06) 0.79 (0.01) 0.51 (0.02) -.37(0.02 ) 0.17 (0.02)

ADG 0.90 (0.04) 0.30 (0.06) 0.62 (0.01) -.47(0.02 ) 0.20 (0.02)

DFI 0.59 (0.12) 0.57 (0.12) 0.24(0.06 ) 0.39 (0.02) 0.37 (0.02)

FC -.42 (0.13) -.58 (0.11) 0.33 (0.17) 0.31 (0.07) 0.14 (0.02)

UB 0.22 (0.13) 0.19 (0.13) 0.39 (0.15) 0.18 (0.15) 0.37(0.07 )

Table4 .Heritabilit y estimates (diagonal) and genetic (lower triangle) andphenotypi c(uppe rtriangle )correlation so fexterio rfactors .

1 2 3 4 5

1.Sid evie wleg s 0.20 -.06 0.05 0.04 -.16 2.Movements/longevit y -.27 0.13 -.02 0.03 0.26 3.Siz e 0.16 0.11 0.39 -.20 -.03 4.Harmon yo fth efram e -.19 -.30 -.38 0.24 0.06 5.Rea rleg s 0.22 0.51 0.21 -.04 0.21

rearpaster n (rg-0.41±0.16) .Uneve nsize dclaw scoincid ewit hsmal ltoe s

(rg=0.61+0.19 ;r-0.53) .Muc hswingin go fth ebac kcoincide swit hbo wrea r legs(r g=0.54±0.16), sickl ehocke drea rleg s(r g=0.52±0.17 )an da sluggis h gait(r g--0.43±0.19 ;r p=-0.40).A lon glifetim eexpectatio ncoincide swit h straight and steeprea r legs (rg=-0.56±0.1 5an d -0.49±0.17), wea krea r pastern (rg=0.57 ±0.15), large toe size (rg-0.53±0.17 ; rp=0.42),smoot h gait (rg=0.46 ±0.16; rp-0.73)an d little swinging of thebac k (rg=-0.51 ±0.17; r_=-0.50). Correlations between performance traits and exterior factorsan dtrait sar epresente di nTabl e6 .

47 CM Csl 01 •tf CO <»H» r- o r~ H « et •0 CM M O vol CM r-l CM CM ' en • n «o CM en •* r» io '-'i M) C 00 O i-l H rH O -tfO-NCM CO (0 CO rH O CM O co|rM m i-i -1 ' CM i-l CM 'H •? CM CM 'T •-'lin

io r-i n CM CM rH ^•I-^I-I >n«o«Nli1 co|(r) «0 ' CM •-I •-, "-1 'CM CM M CM CM en a. I a. i-icoin in cj fv N o> oi en m| in en 3 I I rH . . (r) 10 -'I io

rH VO m VO VO VO f*. CM ~H/ oil H oio-j 1(0 CO CO CM a Ov «l\0H CM m 00 CM CM O i r-l • r-l fr, I CO rH CM C a> CO ^ ^ CM CM m «m o\c» »| H 0| oo H o\ o r* w CM ^, CO CM co r- vo CM ' ' CM CM "I CO r-l m

r- io -3-|-tf in -3- •ï w e» r-l m o -* rH Cv ^ 10 10

00 CO in m CO CO VO M| . o a\ i CM CM

CM r-l| O CM O

aci vol (M CO I CM VO oo If» tH I CO CM •n C f CM r-l CO -3- co vo m m oo m CM 10 i r-l CO r-i 1

m VO vO m in r-l m C r-l r» co r-. H r-* m r-l m r-l r-1 r-l en co 10 CM CM CM CM CM o CM 1 M) « 4-1 a 00 co r-4 m co r~ O O 00 in in r^- rH r-l °l rH rH r~t CM CM CM i r-l CM rH •o Ht col i i CD U .c 4-1 -* CM » CM co C\ ON CO in CM 00 f^ CM r-l CO u CM "1 1-1 co 00 rH co CO CM CM r-l CM r-l r-l CM U i i C o o •ri m oo|t» rH in SO CO a» co m vo vo CM in CM rH rH X P..C E <4H X ,Q M U ai ai o ai tti r-l öS» ? 3 ? 4J CO N U ai ai ai asi ai cd ai bo •H ai >o bo ai •H U •ri TH •H •ri C ai to N £ C S l-l U •o 'H j: > > •O -ri i-H •H •H -ri ai O • „.c > > > > •ri 4-1 O to bO 4-1 4J X Ü oO 4-> 60 4-1 o ai ai 3 4J •rH 4J C ai C ai ai u u C -o C "O W •o •o •O tr o

48 Table6 .Geneti c andphenotypl c correlations (*100)betwee nexterio ran d performancetraits .

Genetic correlations Phenotypiccorrelation s

Exteriorfactor s W/AAD GDF I FC UB W/AAD GDF I FC UB 1.Sid evie wleg s -4-1 0 4 11 -4 0-11 3 4 2.Movements/longevit y -9-1 3 -1 8 6 -5 -5 -3 3 10 3.Siz e 39 48 10 -55 12 17 17 12 -8 4 4.Harmon yo fth efram e 7 9 30 25 49 -6 -4 0 6 8 5.Rea rleg s -23-2 4 0 28 33 -6 -6 -1 6 10

Exteriortrait s 1.Shoulde rblad e 12 24 27 5 51 Back and loin 2. Length 23 5 11 3 3 0 -3 -2 2 -18 3.Widt h 29 41 21 -29 26 18 18 15 -6 14 4.Regularit y 28 29 37 6 47 -2 1 3 2 11 Hams 5.Lengt h 40 41 0 -55 5 1 1 0 -3 -6 6.Widt h 24 34 6 -39 5 16 16 11 -9 -1 Forelegs 7.Fron tvie wleg s 6 13 34 21 10 -1 -3 -1 1 -1 8.Sid evie wleg s 11 14 30 23 8 5 5 4 -3 -2 9.Sid evie wpaster n 44 41 39 -15 14 4 2 2 0 6 Rear legs 10.Rea rvie wleg s 29 -26 27 71 4 -3 -3 -3 2 -4 11.Sid evie wleg s 8 3 -3 -8 -16 4 3 1 -2 -5 12.Sid evie whoc kjoin t 46 -53 -33 23 -23 -3 -2 -1 2 -1 13.Sid evie wpaster n •13 -18 -11 3 6 -7 -7 -2 5 7 14.Liqui da thoc kjoin t •46 -51 -15 37 38 -8 -8 -2 7 6 15.Knottines so fleg s -3 3 13 18 7 -2 -1 0 1 11 Claws rear leg 16.Rati osiz eo fclaw s 3 -6 -28 -21 -25 0 -1 1 3 -2 17.To esiz e -3 -6 5 6 -24 -3 -3 -2 2 -2

18.Gait ;movement s -32-3 5 -7 23 26 -8 -7-5 4 8 19.Swingin go fbac k 9 12-1 3-2 9-1 1 0 10-2 -10 20.Lifetim eexpectatio n 3-2 6 5 11 -5 -4 -3 2 10

W/Aan dAD Ghav esimila rcorrelation swit hexterior ,whic hi sno tsurprisin g becauseo fth ehig hcorrelation sbetwee nbot hperformanc etrait s(Tabl e3) . Genetic correlations are found between daily gain (W/A and ADG) and

exterior factor 'size' (rg-0.39an d 0.48), FCan d 'size' (rg--0.55),an d betweenU Ban d 'harmonyo fframe '( r—0.49) .Geneti ccorrelation sar eals o foundbetwee ndail ygai n(W/ Aan dADG )an dth eexterio rtrait ssid evie w

49 forelegpaster n (rg=0.44±0.1 2 andr g-0.41±0.13) , sidevie w ofhoc k joint

(rg—0.4 6 ±0.14 and rg—0.5 3 ±0.15) and dryness of hock joint (rg—0.46

±0.16 and rg--0.51 ±0.16). Slower growing animals have steeper foreleg pastern, less liquid at the hock joint and more sickle hocked rear legs. Genetic correlations of length and width with daily gain indicate that faster growing animals aremor e likely tohav e abroa d back and loins and broadan dlon gham s atth esam eweight .A hig hgeneti ccorrelatio n isfoun d between FC and rear view rear legs (rg-0.71 ±0.15); boars with better FC have more 0-shaped rear legs. Animals with loose shoulder blades and irregular back and loins have thinner backfat. Low correlations between performance traits and lifetime expectationwer e found. Correlationsbetwee nexterio rfactor san dtrait sar epresente di nTabl e 7. Phenotypiccorrelation sar ei nharmon ywit hth eregressio ncoefficient sfo r the exterior factors (Appendix). High genetic correlations are mostly associatedwit hhig hphenotypi ccorrelations .Exterio r factor 'movements/ - longevity' alsoha shig h genetic correlationswit hsid evie w rear legs and sidevie whoc kjoint .'Size 'i si nadditio nt olengt han dwidth ,genetical ­ ly also correlated with side view foreleg, side view foreleg pastern and rearvie w rear legs. The factor 'rear legs' has genetic correlations with length, rear legs, gait and lifetime expectation.

DISCUSSION

Heritability estimates of linearly scored exterior traits are in the samerang e asestimate s of linearly scoredexterio r traits indair y cattle (Thompson et al., 1983; Diers and Swalve, 1987). Effective selection on most of the linearly scored exterior traits will be possible. Genetic parameters of performance traits are in good agreement with Sönnichsen et al. (1984a,b). Heritability estimates of performance traits are some higher than those reported by Johansson et al. (1985) and Merks (1987). Those literature estimates were based on data obtained under a restricted feeding regime which could be an explanation for the lower estimates. Genetic correlation between average daily gain and feed conversion (-0.58)i s lower than that reported by Johansson et al. (1985) andMerk s (1987), -0.98 and -1.0,respectively . This might alsob e caused by the difference in feeding regime.

50 Highgeneti ccorrelation s (Table6 )o fsom eexterio rfactor san dtrait s withperformanc e traits indicate thatselectio no nperformanc ewil lals o change the animal's exterior and vice versa. Low genetic correlations indicatetha tn olinea rgeneti crelationshi pexist sbetwee ntraits ;howe ­ vernon-linea rrelationship smigh tstil lb epossible ,especiall yfo rtrait s wherebot hlo wan dhig hscore sar eindication so fpathologica ldisorders . Mostle gtrait sar edefine dsuc htha tbot hhig han dlo wscore sar ethough t tob eundesirable .Therefor ei flarg egeneti ccorrelation sbetwee nle gan d performancetrait sexist ,the nth esig no fth ecorrelatio nan dth epopulatio n

Table7 • Geneti can dphenotypi ccorrelation s(*100 )betwee nexterio rtrait s andexterio rfactors 1.

Geneticcorrelation s Phenotypiccorrelation s Exteriortrait s 1 2 3 4 5 1 2 3 4 5 1.Shoulde rblad e -7 -6 -10 78 26 5 24 -7 77 18 Back and loin 2.Lengt h 11 -29 -47 36 -49 -8 -11 -21 36 -5 3.Widt h 4 -6 87 -8 22 7 -1 74 4 4 4.Regularit y -33 -9 -7 83 1 0 24 1 78 19 Hams 5. Length -12-2 4-1 4 13-4 5 -9 -1 42 21 -3 6.Widt h 18 18 94-3 9 26 11-1 3 81-2 1-1 0 Forelegs 1. Frontvie wleg s -41 5-2 3 13 -7 6 4-2 2-2 8.Sid evie wleg s -22-3 1-4 8 38-4 8 9 -19 -9-1 2-2 3 9.Sid evie wpaster n 52 5 65-4 2 22 53 10 24-1 0 11 .Rear legs 10.Rea rvie wleg s 23 -4 -51 30 12 7 -21 -38 25 3 11.Sid evie wleg s 68 -70 7 0 -30 61 -38 5 -1 -26 12.Sid evie whoc k 67 -52 -29 2 -6 70 -10 -4 12 -20 13.Sid evie wpaster n 51 55 2 -23 52 55 31 8 2 13 14.Liqui di nhoc k 21 37 -15 13 71 -12 17 -10 25 80 15.Knottines so fleg s 30 29 20 7 50 -16 19 -4 14 75 Claws rear leg 16.Rati osiz eo fclaw s 33 11 -5 -39 19 1 45 8 -14 33 17.To esiz e 31 42 16 -39 50 22 52 14 -15 26

18.Gait ;movement s -17 63 15 -3 51 -4 79 -6 15 33 19.Swingin go fbac k 46-6 8 20 -3-1 8 9 -69 11-1 6-1 9 20.Lifetim eexpectatio n -16 91 15-2 5 51 -4 86 -2 19 40

1 1-Sid evie wlegs ; 2-Movements/longevity ; 3-Siz e 4-Harmon yo fth eframe ; 5-Rea rleg s

51 meano fth ele gtrai tdetermin ewhethe rth eshor tter mrespons ewil l result ina desirabl e ora nundesirabl e changeo fth ele gtrait .O nth elon g term, whencorrelation swil l continue tob e stable,th ecorrelate d response will alwaysb eundesirabl e and therefore attentionha s tob epai d to that trait (e.g. independent culling or restricted selection index). When selection isbase d ona now nperformanc e indexwit h average daily gain, daily feed intake and backfat thickness as index and breeding goal traits (genetic and phenotypic parameters from this study and economic valuesfo raverag edail ygain ,dail yfee dintak ean dbackfa tthicknes s from Knap etal . (1985)), thensevera lexterio r traitswil lchang e considerably asca nb e seeni nTabl e 8.Boar swil lgro wbroade ra t10 5kg ,th erea rvie w rearleg swil lchang e tomor e0-shape , thesid evie wo fth ehoc kjoin twil l become more steep, more liquid will be present at the joint and the gait patternwil l slow down. Inth emateria l analyzed inthi sstud yn ooveral l legweaknes s score as such was given. Relationships of leg weakness with exterior traits and performance therefore couldno tb e estimated. Inliterature ,however ,gai t patternusuall y isa majo relemen ti nth edeterminatio no fth ele gweaknes s score (Gr^ndalen, 1974; Lundeheim, 1987). When considering leg weakness and gait (movements) score tob e analogous, itca nb e concluded (Table 6) thatselectio nagains t legweaknes swil l influence growth,fee d conversion and backfat thickness in an undesirable direction. The antagonistic relationshipbetwee nle gweaknes san dgrowt hi si nagreemen twit h Lundeheim (1987). However, Bereskin (1979) and Sather and Fredeen (1982) found the associationt ob ebeneficial .Th egeneti ccorrelatio nbetwee n legweaknes s and UB agrees with literature (e.g. Webb et al., 1983; Lundeheim, 1987; Rothschild et al., 1988). Total genetic response andgeneti c change ingai t (movements)hav e been calculated in a situation with two-stage index selection. In the first stage boars are selected on gait pattern only. In the second stage boars are selected on an index with ADG, DFI, UB and gait pattern as index and breeding goal traits.Calculation sar ebase do nparameter s from this study while economic values of performance traits are adopted from Knap et al. (1985). Economic value of gait (movements) isse t to zero.Geneti c change in gait pattern and total genetic response is calculated for varying fractions of boars selected in the first stage while total selected

52 fractionafte r two stageswa skep tconstan ta t0.0 5 and ispresente d in Figure1 .I nth epresen tDutc hherdboo ksituatio napproximatel y50 %o fal l testedboar sar erejecte dfo rbreedin gbecaus eo fle gweakness .I tca nb e seenfro mFigur e 1tha t intha tsituatio ngai tpatter nwil lchang e ina n undesireddirectio nan dtota lgeneti crespons ewil lb e76 %o fwha twil lb e achievedwhe nselectio noccur sonl yi nth esecon dstage .Whe nrejectin g63 % ofth eboar si nth efirs tstag ethe ngai tpatter nwil lno tchang ean dtota l geneticrespons ewil lb e64 %o fwha twil lb eachieve dwhe nselectio noccur s onlyi nth esecon dstage .Whe nboar sar eselecte donl yi nth efirs tstag e thentota lrespons ewil lb enegative .

Table8 .Geneti cchang ei nexterio rtrait swhe nselectin gfo rperformance .

Exteriortrait s 5G ^1i,index Yi 1.Shoulde rblad e .0038 .0140 Back and loin 2.Lengt h -.0014 -.0052 3.Widt h .0316 .1168 4.Regularit y .0049 .0181 Hams 5.Lengt h .0187 .0691 6.Widt h .0341 .1260 Forelegs 7.Fron tvie wleg s -.0032 -.0118 8.Sid evie wleg s -.0017 -.0063 9.Sid evie wpaster n .0187 .0691 Rear legs 10.Rea rvie wleg s -.0366 -.1352 11.Sid evie wleg s .0052 .0192 12.Sid evie whoc kjoin t -.0301 -.1112 13.Sid evie wpaster n -.0125 -.0462 14.Liqui da thoc kjoin t -.0254 -.0938 15.Knottines so fleg s -.0009 -.0033 Claws rear leg 16.Rati osiz eo fclaw s .0065 .0240 17.To esiz e -.0053 -.0196

18.Gait ;movement s -.0222 -.0820 19.Swingin go fbac k .0136 .0502 20.Lifetim eexpectatio n -.0046 -.0170

1 regressiono fexterio r trait YLo ninde x 2 change in exterior trait Yi as a result of one round of selection (selectionintensity :1 )o nth einde x

53 too• )

10 20 30 40 50 60 70 tO tO 100 Fraction («100) »Itetid In flr»t «tog«

Figure 1.Geneti c change in gait (- -)an d total response (• -)fro m a twostag e selection indexwit hvaryin g fractions selected inth e first stage and a total selected fraction of 0.05.

When selection decisions have to be made, it is difficult to use information of a large number of separate exterior traits. Combining information onal ljudge d exterior traits intoa fe w factors canb e anai d in handling that problem. In this study moderate heritability estimates were found for five analyzed exterior factors, which indicates the possibility ofeffectiv e selectionfo rexterio rb y those factors.However , relationships between exterior factors and performance traits are more difficult tointerpret .Reductio no fth enumbe ro fexterio rtrait so nwhic h a selection decisionha s tob ebase d canals ob e achieved by judgingjus t a few relevant exterior traits. The choice which exterior traits have to bejudge dmus tb ebase do n their economicvalu e andheritability . However, no information is available in literature about the economic values of exterior traits.Th e economic values depend among others on the relation-

54 shipwit h sowproductivit y and longevity,whic h issubjec to f study intw o subsequent chapters.

ACKNOWLEDGEMENT

Part of the costs of this study were coveredb y financial support from the Dutch Commodity Board for Livestock andMeat .

REFERENCES

Bereskin, B., 1979. Genetic aspects of feet and legs soundness in swine. J. Anim. Sei. 48:1322-1328 . C.B.V., 1985.Annua l report of aboa r test station (Jaarverslag selectie mesterij Zuid-Nederland, 1985). Diers, H. and Swalve, H., 1987. Estimation of genetic parameters and breeding values for linear scored type traits. 38th Annual Meeting of the EAAP, Portugal, abstract G5a20. Grondalen, T., 1974. Leg weakness in pigs. I. Incidence and relationship to skeletal lesions, feed level, protein and mineral supply, exercise and exterior conformation. Acta Vet. Scand. 15:555-573 . Hill, W.G., 1984. On selection among groups with heterogeneous variance. Anim. Prod. 39:473-477 . Johansson,K. ,Andersson ,K .an dLundeheim ,N .,1985 .Evaluatio no fstatio n testing in pigs. I. Genetic parameters for feed measurements and selectioneffect so nvoluntar y feedintake .In :K .Johansson ,Estimatio n ofgeneti cparameter s forus ei nth eSwedis hpi gbreedin gprogram ,repor t 66 of the Swedish Univ. Agr. Sei., Dep.Anim . Breed. Gen.,Uppsala . Kennedy, B.W., 1981. Variance component estimation and prediction of breeding values. Can.J . Genet. Cytol. 23:565-578 . Kennedy, B.W., Johansson, K. and Hudson,G.F.S. , 1985.Heritabilitie s and genetic correlations forbackfa t andag ea t9 0k g inperformanc e tested pigs. J. Anim. Sei. 61: 78-82. Knap,P.W. , Huiskes,J.H . andKanis ,E. ,1985 .Selectio n index for central test in Dutch pig herdbook breeding from 1984.Livest . Prod. Sei. 12: 85-90.

55 Lundeheim,N .,1987 .Geneti canalysi so fosteochondrosi san dle gweaknes s in theSwedis h PigProgen yTestin g Scheme.Act aAgric . Scand. 37:159-173 . Merks, J.W.M., 1987. Genotype x environment interaction in pig breeding programmes: II.Environmenta l effects andgeneti cparameter s incentra l test. Livest. Prod. Sei. 16:215-228 . Meyer,K. , 1985.Maximu m Likelihood estimation ofvarianc e components for a multivariate mixed model with equal design matrices. Biometrics 41: 153-165. Meyer, K., 1986. Restricted maximum likelihood to estimate genetic parameters - In practice. 3rd World Congress Gen. Appl. to Livest. Prod.,Nebrask aXII :454-459 . Meyer,K. , 1987.A note on the use of an equivalent model to account for relationships between animals in estimating variance components. J. Anim. Breedg. Genet. 104:163-168 . Reiland, S.,Ordell ,N. ,Lundeheim ,N .an dOlsson ,S.E. ,1978 .Heredit y of osteochondrosis, body constitution and leg weakness in the pig. A correlative investigation using progeny testing. Acta Radiol, suppl. 358: 123-137. Rothschild, M.F. and Christian, L.L., 1988.Geneti c control of front leg weakness in Duroc swine. I. Direct response to five generations of divergent selection. Livest. Prod. Sei. 19:459-471 . Rothschild,M.F. ,Christian ,L.L . andJung ,Y.C. ,1988 .Geneti c control of front leg weakness in Duroc Swine. II. Correlated response in growth rate, backfat, and reproduction from five generations of divergent selection. Livest.Prod . Sei. 19:473-485 . Sather,A.P . andFredeen ,H.T .1982 .Th eeffec to fconfinemen thousin gupo n theincidenc eo fle gweaknes s inswine .Can .J .Anim . Sei. 62:1119-112 8 Smith,C. ,1966 .A not e onth eheritabilit yo fle gweaknes s scores inpigs . Anim. Prod. 8:345-348 . Smith, R.G.C. and Smith, C., 1965. Report onboa r performance testing at Stirling. Anim. Prod. 7:284 . Sönnichsen, M.L., Claus, J. and Kalm, E., 1984a. Genetic parameters and selection indicesfo rLandrace- Ban dPletrai ni nSchleswig-Holstein . 1. Estimation ofheritabilities . Züchtungskunde 56:238-248 .

56 Sönnichsen, M.L., Claus, J. and Kalm, E., 1984b. Genetic parameters and selection indices forLandrace- B andPietrai ni nSchleswig-Holstein . 2. Estimation of covariances. Züchtungskunde 56:249-261 . Swanson, G.J.T. and Bryan, G.J., 1984.Linea r assessment of conformation inFriesia nHolstei ncattle .35t hAnnua lMeetin go fth eEAAP ,Th eHague , abstractG5.36 . Teuscher,T. ,Weniger ,J.H . and Steinhauf,D. , 1972.D eBeinschwäch e beim Schwein. Züchterische und produktionstechnische Bedeutung. Schweinez. Schweinemast 20:292-296 . Thompson, J.R., Freeman, A.E., Wilson, D.J., Chapman, CA. and Berger, P.J., 1981. Evaluation of a linear type program inHolstein s J. Dairy Sei. 64:1610-1617 . Thompson,J.R. , Lee,K.L . and Freeman,A.E. ,1983 .Evaluatio n of a linear type appraisal system for Holstein cattle.J . Dairy Sei. 66:325-331 . VanSteenbergen ,E.J ., 1989 .Descriptio nan devaluatio no fa linea rscorin g system for exterior traits inpigs .Livest . Prod. Sei. 23:163-181 . Webb, A.J., Rüssel, W.S. and Sales,D.I. , 1983. Genetics of leg weakness inperformanc e tested boars.Anim . Prod. 36:117-130 .

57 CO CM rH o CO co VO co co rH CV in 4-1 in O o o o o I-I o O O o m 1-1 I-l o O o G o o o o> o 1 o o 1 o l l o o 1 1 o o O O o 1 O 1

co ui 00 CO VO VO CM o o VO o-d- in OH <* .-1 O o O rH o rH O rH O i-I CM O

•d vo in CM co I-I i-i 10 l-~ 00 in CM CO o CO CM co VO CO 1-1 ai rH O o o O 1-1 o o o CM o CO CO co o o O o O O o O 1 o o 1 l o o o o O o o 1 1 O o 1 1 1 TJ u co •O C

•o X X C O O ta ai O CU •H •o rH T) Xi rH 3 r-l 01 -O bil 01 bf bf] a- a o. ai G (X 0) ai C a> o> •o Tl rH T) cO r-l CO ot rH rH CO rH 0) •r-l M 10 3 M a A rü x: X 4J e 00 >,x.- bü rH c O bil e o CO u 3 01 3 O 3 4-1 o rH U 01 Cl C O o rJ m o r-l • 3 O 1 o •H O •H CO cO 3 O 01 rH XI u 1-1 XI |x! Xi rH XI XI tl) rH rH o Al i-i ai K rH

a> 4-1 u 'S ü o X •H VO ^^ o U ai 0) CO 4-1 4J rH rH co u 01 bO bO •H o bfl 4J n CO rH (X co •s•H . O. O. CM rH 4-1 0) U u 01 I-l i-i .C rK! 01 X. cO 01 01 x; CM ~^rH 3 u O n u U n i-i 01 CJ 01 01 U 01 01 tl rH rH CO o rH o O rC m U rC co t -H 4-1 i 4-1 4-1 u 3 e rH •H x; c r-l co e •H 01 C O oi 01 O 01 01 tl) B Al VI til 01 e 01

o rt 4-1 4J c VJ C c 01 o m O •H •H 0) 3 •H o 4-1 c O e O on cO 4-1 t) u 1-1 V4 1—) 01 rH .ü cO C cO co 01 01 rH O 01 ü 4-1 o H-l ai bO 01 4J 01 0] Ai 4J A! 4J co O •H x> H Xi X 3 CO 01 01 o 0) •M 4J 0) 3 3 3 3 3 3 4J 0) N ^Cl IJ •H .H 01 01 ai 01 01 ai CO 01 bO •H 01 o> bO 0) 01 U Ol c VJ •H •H •H •H •H •H e co 01 N e e e 0) O •o •H X ca X! > TJ *H r-H •H •H •H a m rH O 4-1 X! rH *-> x; 4-1 > > > > > > •H 4-1 O 01 bO 4-1 3 r-H bO 4-1 3 bO 4J c 01 01 i-i 01 ai 01 3 4-1 U •H 4-1 C 01 O C TJ bO C T3 O •o •o 0) cO TJ TJ TJ er O 4J 01 •H •H 4-1 X X! •a ai «H ai 01 >H 01 u •H •H CX> 01 •H •H •H •H c «01 CO O CO 3 •H •H C/3 c •J S ei rJ 3 bnPu CO H CA H O co rJ T) co ai « -H « C -H 0) CU rH CM CO a; fc, tk! vj

58 Chapter 4

EXTERIOR IN SOWS: 1.EFFEC T OF PARITY NUMBER AND ASSOCIATION WITH REASONS FOR DISPOSAL

E.J. van Steenbergen, E.Kanis ,W.J . Koops andH.A.M .va n der Steen

Department ofAnima l Breeding, WageningenAgricultura l University, P.O. Box 338,670 0A HWageningen , TheNetherlands .

Submitted for publication inLivestoc k ProductionScience . Reproduced by permission of Elsevier Science Publishers B.V., Amsterdam.

59 ABSTRACT

Over a two-year period, several thousand gilts between 7 and 9 months of age and sows from weaning till about 30 days post weaning, were repeatedly judged for exterior on 24commercia l multiplier herds.A total of 20 linearly scored exterior traits (especially those concerning legs) wereconsidered . Exterior traitso fgilt san dsow swer edescribe dan dasso ­ ciation of exterior with reasons for disposal was investigated. Lengtho fth eanimal s increasedunti l6t hparit ywherea swidt h increased over the 10parities . Pasterns of fore- and rear legsbecam eweake r after 5thparity .Th edevelopmen to fsiz etraits, lengt ho fback ,loin san dhams , and toe size indicated thatmaturit y wasprobabl y reached at 6thparity . Principal component analysiswa suse d toindicat erelationship s between exterior traits and tomar k therelevan texterio rvariable s inrelatio n to reasons for disposal. Some exterior traits of gilts,showe d a significant relationship with reasons for disposal, especially leg weakness. Sows culled because of leg weakness before 4th parity were longer and broader as gilts,ha d less straight rear legsan dmor e tubercles at the rear legs, theywalke d sloweran dtwiste dmor ewit hth ehin dquarter scompare d tosow s thatha d produced at least four litters.Sow s culledbefor e fourth parity becauseo fle gweakness ,ha da lowe rscor efo rlifetim eexpectatio n (number oflitter s expected still tob eproduced )a ta ag eo f7- 9 months than sows thatproduce d > 4 litters.

INTRODUCTION

In the literature, studies on exterior inpig s have been concentrated onbod y size andbon e characteristics offore - andrea r legs (e.g., Hetze r and Miller, 1972; Grandalen, 1974; Van der Wal et al., 1987). The relevance of exterior traits in pigs depends on their relationship with involuntary culling and on their predictive value for other economically important traits. Involuntary culling reflects problems, e.g., leg weakness, diseases, and infertility, that are manifested in premature disposal. Advantages of decreased involuntary culling are: (1) reduction in annual cost of replacement, (2) increase of average number of piglets perlitte rbecaus eo fdecreasin gnumbe ro ffirs tparit ysows , (3)reductio n

61 in number of non-productive days, and (4) increased opportunities for selectiono nothe r traits. Incommercia lherds , exterior of sows is mainly used to predict the risk for involuntary culling. Relationships between exterior traits in pigs and involuntary culling were not found in literature. In dairy cattle, Rogers and McDaniel (1989) mentioned consistent relationships of the type traits,udde r depth, teat placement, and foot angle,wit h culling. Theobjectiv e ofthi s studywas ,t ocharacteriz e exterior of ananimal , to describe the change in exterior traits with parity number, and to estimate relationships between exterior and reasons for disposal.

MATERIAL

In 1985 and 1986, gilts and sows were judged for exterior at 24 commercial multiplier herds with 100-200 sows. Farms participating in the project obtained feed from the same company, and sow information was registered by a computer Sow Management System, supplied by this feed company. Farmswer evisite d regularlyb y theextensio n serviceo f the feed company. Therefore differences between farms due to feed and/or feeding regimewer e assumed not tob e relevant. A total of eleven inspectors of the Dutch herdbook and three breeding companies rotated over the farms to score exterior of the animals. In total, 20 exterior traits (Appendix) were scored on a linear scale (Van Steenbergen, 1989). Farm s were visited monthly by an inspector who judged allgilt s ofabou t 7-9 months ofag e andal l sows fromweanin g (on average 28 days after farrowing) till about 30 days post weaning. So most of the giltswer e judged twice by different inspectors and most of the sows were judged once after each weaning of thepiglets .Ove r aperio d of two years 6333animal swer ejudge donc e ormore .Combinin g exterior informationwit h sow productivity information reduced the number of animals with both exterior and production records to 5804. Data were further checked for following criteria: i. giltsha d tob e judgedbetwee n 150an d 300 days of agean dii .sow sha d tob ejudge dbetwee nweanin g and4 0day spos tweaning , while the parity number was restricted to a maximum of 10. After this screening, data included 5660animal s onwhic h 16456judgement s weremade .

62 The distribution of number of judgements per animal in relation to paritynumbe ro ffirs tjudgemen t isshow ni nTabl e1 .O fal ljudgements , 91% wasbefor e parity 6; approximately 2%o f thejudgement s made after weaningwer e repeated appraisals inth e sameparity ; 502o f theanimal s (2906)wer ejudge dbetwee n15 0an d30 0day so fage .

Table1 .Distributio no fanimal sove rnumbe ro fjudgement spe ranima lan d theparit ynumbe rwhe nfirs tjudged .

Parity number Numbero f judgement.3 pe r animal whenf ] Lrstjudge d 1 2 3 4 >5 Total 0 183 697 918 620 488 2906 1 165 110 124 171 132 702 2 134 136 141 148 89 648 3 95 96 128 75 20 414 4 96 138 87 36 25 382 >5 259 235 71 35 8 608

Thedistributio nove rstrain si si nTabl e2 .On ethir do fth esow swer e crossbredfro mDutc hYorkshir eboa ran dDutc hLandrac esow .O nmos tfarms , animalswer emainl yfro mon estrain .

Table2 .Distributio no fanimal sove rstrains .

Strain %

DutchLandrac e (DL) 10 DutchYorkshir e(DY ) 3 DL* D Y 10 DY* D L 32 Duroc* D L 11 Hypor1 15 Other 19

:Commercia lcrossbre dsow .

63 During the experiment over 70%o f the animals were culled for various reasons. Farmers could record only themai n culling reason out of eight reasons indicatedb y theso wmanagemen t system. During onean da hal f year a form was send to the farmers on which they could record: (1) culling reasons inmor e detail and (2)a second andthir d culling reason in order of relevance.

METHODS

Exterior scores were transformed to a logit scale (Bliss, 1970) to overcome twoundesire d aspects ofth e scoring system: (1)Exterio r traits have been scored on a linear scale from zero to nine with 0.5 intervals. When scoring has not taken place at all parities, extrapolationmigh tresul t inth eundesirabl e situationtha tscore so nth e scale of0 to 9ca nfal loutsid e thelimits . (2) Although the scale is defined linearly, it is expected, because of limiting thescal ewithi nth erang eo f0 t o9 ,tha tinspector swil lno tus e equidistant steps over the whole scale. For example steps in exterior scores from 1 to 0 and from 8 to 9 are expected tob e larger than a step from4 to5 .

The general logitdefinitio n is:Logi tY - y '- I n [Y / (Yu -Y ) ]wit h lower limit of zero anduppe r limitY u.Whe nchangin g the formulato : y' •=I n [(Y+ b) / (Yu +b -Y )] wher eb is aconstan t (b>0), al l scores including0 an d9 ca nb e transformed tologi tscale .Th e logittransforma ­ tionwa sperforme do nal lexterio rscore swit hb arbitrar yse tt o0.1 .Bac k transformation toorigina l scale canb e doneby :

Y. i - ey -b - i + e-y' i+ ey'

Effect of parity on exterior traits was analyzed with the following model, using the LSML76progra m ofHarve y (1977): yijtan- M + HS,+ A li(+ Ik+ Pm+ eiJtan (1) where, yijtan = logit °f the score fora nexterio r trait fi = mean logit score HS, - fixed effect ofherd-strai n group i (i-1 74)

64 A^ - random effect of animal jwithi nherd-strai ni

Ik - fixed effect of inspector k (k-1 11)

Pm - fixed effect ofparit y m (m-0 10) eljkmn - random error term

Least squares means forparit y numberwer euse d to evaluate relationships between exterior traits andparit ynumber . Inorde rt osimplif yanalysi san dinterpretatio no fexterio rtrait s from different parities, an attempt was made to combine all information per animal. Information from gilts (judged at 150-300 days of age),however , was not combined with information obtained on the same animals as sows (judgedfro m0 t o-3 0day spos tweaning )becaus eo fdifference s inhousing , feeding andcondition .Becaus eo fth esignifican t effecto f inspectors and herd-straingroup so nlogi tscores ,al lscore swer ecorrecte dfo r inspector and herd-strain by least squares constants estimated with model 1. For giltswit h two ormor e records,correcte d scoreswer e averaged per trait. Preliminary analyses showed a significant effect of parity number on scores of most exterior traits. Exterior scores were corrected for the effect of parity number and expressed as deviation from the parity mean (DM). Parity means for the exterior traits were estimated per exterior trait by a second degree polynomial fitted to the parity least squares means (model 1),weighte d by their effective numbers. When two or more judgements were made on a sow, DM was averaged per trait. For sows with repeated judgements, the deviation of the linear regression on parity number (DLR) of each of the exterior traits, from the average linear regression was estimated. Orthogonal polynomials (OP) were used to estimated both DM and DLR. OP'swer e fitted to the corrected logit scores with the IMSLroutin eRLFOT W (IMSL, 1984),wit hweighin g coefficient 1fo r parities with an observation and 0 forparitie swit h no observations. DLR was estimatedwhe na sowha dbee njudge dmor e thantw o times and otherwise set tozero .

Thisresulte d ina totalo f6 0exterio rvariable s divided into 3group s of 20 exterior traits each. The first group consisted of traits that referred toexterio rjudge da tgil tstag e (GE), whereas thesecon d (DM)an d third (DLR)grou preferre d toexterio rtrait sjudge do nsows .D Mmanifeste d the deviation from the overall average level and DLR indicated the

65 deviationfro mth eoveral llinea rregressio no fth eexterio rleas tsquare s meanso nparit ynumber . Thederivatio no fth e6 0exterio rvariable sfro mth eorigina lexterio r scoresi ssummarize di nFigur e1 .

Principal component analysis was used to reduce the large number of variables toa smalle rse to f independentfactor s inwhic h thevariable s were grouped, and to study the relationships between the 60variables . Principalcomponen tanalysi si sa thre este pprocedure :(1 )th ecomputatio n of thecorrelatio nmatrix , (2)th e extractiono f the initialorthogona l factors,an d (3)th erotatio nt oa termina l solution.Initia lorthogona l factors composed of combinations of traits, based on the correlation structurebetwee nthos etraits .Th econtributio no fa trai tt oa facto rca n be changed by factor rotation.Whe n rotating factors, a discrimination betweentrait s issought ,wher eonl ya fe whighl ycorrelate d traitsmak e a large contribution to the factor,an d the remaining traits contribute onlymarginally .Thi ssimplifie sth einterpretatio no fa factor .Th enumbe r of factors to retain is arbitrary,bu t isbase d mostly on theireigen ­ value, i.e., thevarianc e explainedb y each factor.Fo rG E traits,275 4 animals had missing values because they were not judged at age of 7-9 months.T oanalyz edat ao fal l566 0animal stogether ,leas tsquare smean s forth eanimal' sherd-strai ngrou pwer eallotte dt omissin gvalues .I nthi s study factors were extracted with principal component analysis by the FACTORprocedur eo fSA S(SA SInstitut eInc. ,1985) .Orthogona lrotatio nwa s doneb yth eVARIMA Xoption . Resulting factorswer e related tocullin g reasons.B y combining some culling reasons mentioned in Table 3, the following five reasons were considered: (1)le gweakness , (2)udde r problems, (3)productivity , (4) reproductionfailure s and (5)unknown .I twa sassume d thatsow s thatha d produced atleas tfou r litterscoul db eclassifie da sanimal stha tha da satisfactory constitution. Factor scores of these animals were compared with scores of animals culled (forvariou s reasons)befor e 4th parity. Linear contrastsbetwee nnon-culle danimal san d thoseculle dfo rvariou s reasonswer eestimate d by theGL Mprocedur e ofSA S (SAS InstituteInc. , 1985)wit hth efollowin gmodel :

YiJk =p + HS i+ CR j+ e ijk (2)

66 GILTS SOWS LSM forparity , Inspector and herd/strain.

Average scores Deviations from per animal estimated parity means

Zero or first degree orthogonal polynomial

20 exterior traits ingrou p sow level deviation. 20exterio r traits ingrou p sow course deviation.

15 exterior factors.

Figure 1. Scheme of transformation of exterior scores to final exterior variables (see text for explanation).

67 where,Y 1Jlci s an exterior factor score, n the overall mean, HSj^th eherd - straingrou p i (i-1,. . ,74 ), CR jth ecullin greaso nj (j-1 6;no tculle d orculle d forvariou s reasons),an de lJk therando merro r term.A n exterior indication for risk on premature disposal, obtained at an early age, can be an interesting tool to decrease premature disposal. Exterior traits judged at gilt stage were therefore also studiedwit h model 2.

RESULTS

Distribution of animals over culling reasons as reported in the sow management system isgive n inTabl e 3.Almos t one third of theanimal s was culledbecaus e ofunsatisfactor y production results (litter size)an d23 % because offertilit yproblems . Legweaknes swa s thecullin g reasonfo r11 % ofth edispose d sows,whic h is inclos e agreementwit h figuresreviewe db y Pattison (1980).Detaile d reasonsfo rdisposa lwer eavailabl e for89 8sows . Amajo ran da mino r culling reasonwer egive nfo r25 %o fth esow san dthre e culling reasons were given for 3%. InTabl e 3 frequency distribution over the first mentioned culling reasons is given.Whe n leg weakness could be recorded in more detail, a larger portion of the animals (8%more ) were culled because of that reason, compared to a similar reason in the sow management system. Twenty five of the 131 animals (19%) disposed because of leg and claw disorders had a second culling reason which referred to productivity or reproduction failure. A detailed recording of culling reasons does not seem to increase the reliability of the information. Combined distribution of all exterior scores is in Figure 2.Ther e was preference for 'whole' scores,whic hwa s found also inanothe r experiment by Van Steenbergen (1989). Standard deviations of the 20 exterior traits are similar with results found inboar s (VanSteenberge n et al., 1990). Withmode l (1),4 7 to64 %o fth etota lvarianc e inlogi texterio r scores was explained,wherea s thevarianc e explainedb yparit ynumbe r ranged from 0 to 5%. An effect of parity number was found for all traits (P<0.01), except side view rear legs. Of all exterior traits, the second degree polynomial, fitted to the parity least squares means ofparit y 1 to 10 is presented graphically in Figure 3. The average gilt scores are also indicated in that figure. Back, loin, and hams of gilts are broader than those of sows after the first weaning. There is also an increase in both

68 Table 3. Distribution of disposed animals over culling reasons.

Reasons reported by the SowManagemen t System.

Number X Reason 311 8 Anoestrus 613 15 Failure toconceiv e or tomaintai n pregnancy 452 11 Leg weakness 229 6 Udder problems 192 5 Vitality 1193 30 Productivity (mainly litter size) 323 8 Disease 677 17 Unknown

First (main)cullin g reasons of detailed reporting.

Number X Reason 14 2 Forelegs 131 15 Rear legs 17 2 Claws 65 7 Udder 72 8 Old age 198 22 Productivity (mainly litter size) 40 4 Abortion 81 9 Anoestrus 157 17 Failure tobree d after two matings 50 6 Disease / inflammation 83 9 Other reasons

length and width after first weaning. Front view forelegs and side view pasterno fbot h fore-an drea rleg sshowe da curvilinea r trendwit hparity . From side view forelegs, an increase inbuckle d knees after third parity canb e noticed. After 5th parity, rear legs showed a tendency to be less sickle hocked. Quantity of liquid at the hock joint (swelling) decreased with agewherea s knottiness ofrea r legs increasedwit h age.Siz erati o of claws and toe size showed an increase till 5th parity, and stabilized thereafter. After a stable gait pattern from first to fifth parity, gait sloweddown .Lifetim eexpectatio nscore ssho wa linea rdecreas ewit hparit y

69 number, as expected, because of the definition of the trait (expected numbero flitter sstil lt ob eproduced) .

PERCmTAGE 20

16

12

0O51 1.52 ZS3 0 54 4. 55 5. 56 & 57 7. 580 59 CATEGORY

Figure2 .Overal ldistributio no fexterio rscores .

A factor analysis was performed on the 60 exterior variables. Twenty factors had an eigenvalue >1, which jointly explained 61%o f the total exteriorvariance .Afte rrotatio no fthos e2 0factors ,eigenvalue so fth e factors andcorrelatio nwit h theexterio rvariable swer eexamined .Afte r a gradual decrease of the eigenvalues of factor 1 to 15, a markedly stronger decrease wasnotice d from factor 15 tofacto r 16.Thi swa s the main reason to restrict further analyses to factor 1 to 15. Those 15 factorsexplaine d51 %o fth etota lvariance .Fo reac hfactor ,th evariable s witha correlatio nwit hth efacto rgreate rtha n0.50 ,ar ei nTabl e4 . Deviation between exterior factor scores of non-culled animals and animalsculle dfo rvariou sreason s (contrastestimates )ar ei nTabl e5 .A significant effect ofexterio r oncullin g for legweaknes s isfoun d for factors 1, 2, 4, and 13. Those factors refer to exterior of gilts, particularly their width, rear legs and claws, locomotion and lifetime expectationscore ;an d toexterio ro fsows ,th edeviatio no f theaverag e (DMgroup )fo rsid evie wrea rlegs ,gait ,swingin go fth ebac kan dlifetim e expectation score.Exterio r of animals culled for reproduction failures

70 I.Shoulder blad« 2.Length of back and loin 3.Width of back and loin 4.Regularity of back and loin

1.25.

I '' ' ' I '' ' i I M '' ' I' | i i i i I i i i i I | i i i i I i i i i|

6.Widtho fh a 7.Frontvia » forelegs «.Sidavie w forelegs

'I '' ' ' I | I I 1 I | I I I I II I I I I I II | '| II I I | I I I I |

9.Sidevie w pastern forelegs 10.Rearvie w rear legs 11.Sidevie w rear lags 12.Sidevie whoc k Joint

1.25-

. »-. .- 0 I '' ' ' I 'I '' I I I '' ' I I '' ' ' I I' ' ' ' I

17.Toesiz « 18.Gait (movements) 19.Swingingo fbac k 20.Lifetime expectation

1.25-

-1.25-

«•• •e » I I I I I I | i I I I | I I I I | '|' 1 I I I j I I i ^ ] 0 5 10 0 10 0

Figure 3. Lease squares means ofexterio r traits by paritynumbe r and the fie of a 2nd degree polynomial (logit exterior score on the vertical axis and parity number on the horizontal axis).

71 Table4 .Trait swit ha nabs o

Exteriortraits 1

Factor Exterioro fgilt s Leveldeviatio n Coursedeviatio n

1 14t o2 0 2 11,1 8t o2 0 3 18t o2 0 4 3an d6 5 1,2,4 and5 6 1,2 ,4 an d5 7 9,1 1t o1 3 8 16an d1 7 9 3an d6 10 16an d1 7 11 1an d4 12 11t o1 3 13 11an d1 2 14 3an d6 15 9 and1 3

1 :SeeAppendi xfo rexplanatio no fth etrai tnumbers .

before4t hparit yan danimal stha tha dproduce da tleas tfou rlitter sdi d notsho wdifferences .A nassociatio nbetwee nexterio ran ddisposa lbecaus e of udder problems is indicated by the first factor e.g. hind quarter characteristicsa tgil tstage .Analysi so fexterio rtrait sjudge da tgil t stagewa sbase d onlyo ndat ao fanimal stha twer ejudge da sgilt s (Table 6). Most interesting traits are:lengt h ofbac kan d loins (2),widt ho f hams (6),sid e view rear legs (11),knottines s of rear legs (15),gai t (movements)(18) ,swingin go f thebac k (19)an dth elifetim eexpectatio n score (20).Fo rthos etraits ,difference sbetwee nculle dfo rle gweaknes s andnon-culle d animalsvar y from0.1 5 to0.35 ,whic h isabou t 35%o fth e standarddeviatio no fth elogi texterio rscores .Sow stha tha dproduce da t leastfou r litterspossessed ,a sgilts ,smalle rhams ,mor estraigh t legs andfewe rtubercle sa tth erea rlegs .The ycoul dwal keasie ran dquicker .

72 twisted less with theback ,an dwer e expected toproduc e more litters in comparisont osow s thatha dbee nculle dbecaus eo fle gweaknes sbefor e4t h parity.

Table 5.Linea rcontrasts 1o fexterio rfactors ,o fanimal stha tproduce d> 4 litters (H)an d animals culled before 4thparit y for various reasons(L) .

Culling rea sons

Leg Udder Piglet Reproduct. Unknown Factors weakness problems production failures

1 0.28*** 0.32*** 0.22*** 0.05 0.16*** 2 0.41*** -.09 -.01 0.06 0.01 3 0.07 0.11 0.00 0.06 0.07 4 -.30** * 0.04 0.03 -.05 -.05 5 -.01 0.24 0.09 0.08 0.09 6 -.09 -.14 -.07 -.02 -.04 7 -.06 -.10 -.10 0.02 -.13** 8 0.01 0.00 -.07 -.02 0.02 9 0.03 0.07 -.11* -.08 0.00 10 0.06 0.21 0.12** 0.02 0.13** 11 -.02 -.02 0.05 -.02 0.04 12 -.07 -.06 -.06 0.00 0.02 13 -.17* 0.11 0.06 0.09 -.01 14 0.00 -.09 -.05 0.01 0.01 15 0.08 0.06 -.01 0.06 -.03

P< 0.05 ; P< 0.01 ; P< 0.00 1 Contrastestimate da sH minu sL .

DISCUSSION

Manycomponent so fexterio rma yb eexpecte d tochang ewit hincreasin g age, especiallydurin ggrowt ht omaturity .I ndair ycattle ,severa ltyp e traits,e.g. , rear legs,pastern san dudde r traits,ar eaffecte db yag e (Bowden, 1982). Whencullin g isdu et oexterio r traits,the nthi smigh t

73 Table6 .Linea rcontrasts 1o fgil texterio rtraits ,o fanimal stha tproduce d >4 litter s (H)an danimal sculle dbefor e4t hparit yfo rvariou s reasons(L) .

C u 1 ling r e a s ons

Leg Udder Piglet Reproduction Unknown Exter ior traits weakness problems production failures

1. Shoulder blade 0.03 0.11 0.02 0.05 0.03 Back and loin 2. Length -.15*** 0.15 0.01 0.00 -.02 3. Width -.10* 0.06 -.05 -.01 -.05 4. Regularity 0.02 0.05 -.02 0.02 0.01 Hams 5. Length -.03 -.01 -.01 0.00 -.01 6. Width - .20** * -.08 -.06 -.07 -.11** Forelegs 1. Front view legs 0.02 0.14* 0.01 0.01 0.01 8. Side view legs -.04 0.25** -.03 -.01 -.02 9. Side view pastern 0.04 0.10 0.05 0.01 0.02 Rear legs 10. Rear view legs 0.02 0.00 -.01 0.00 0.00 11. Side view legs -.15*** -.08 0.00 0.01 -.05 12. Side view hock joint -.03 0.05 -.01 0.05 -.03 13. Side view pastern 0.01 0.05 -.04 0.04 -.05 14. Liquid at hock joint 0.12 0.12 0.01 -.03 0.06 15. Knottiness of legs 0.18** 0.11 0.07 0.00 0.00 Claws rear leg 16. Ratio size of claws -.10 0.05 0.01 0.02 -.04 17. Toe size 0.06 0.00 0.01 -.02 -.02

18. Gait; movements 0.32*** 0.29 0.10 0.03 0.08 19. Swinging of back -.19*** -.16 -.09 -.01 -.05 20. Lifetime expectation 0.35*** 0.40 0.13 0.01 0.12

Number of animals 154 31 262 315 315

1 :Contras testimate da sH minu sL . have introduceda bia s inparit yestimate so fexterio r traits.Thi sbia s might be more relevant with increasing parity number. Development of exteriortrait si ntim e(Figur e3 )migh ttherefor eb edifferen tfo ranimal s that are culled at an early stage.Increas e of length of sows till6t h parity (Figure 3)migh tb e an indicationfo rmaturit y attha t time.Fo r

74 non-producing sows,Walstr a (1980)foun da matur eag eo fapproximatel y 750 days,whic h is about 1.5 years earlier. With principal component analysis the number of exterior traits for further analysis was reduced from 60 to a more feasible number of 15. The main traits in a factor related to one group (GE, DM or DLR). Relationships between exterior traits within a group were stronger than relationships between the same traits across groups. This might be due partly toth estructur eo fth edataset ,wher emissin gvalue so rvalue s that could not be estimated were replaced by herd-strain least squares means. Correlations between traits within a group were similar for all three groups. Only a few factors showed significant relationships with reasons for disposal. Disposal because of leg weakness was related to three closely correlated traits e.g. gait, swinging of the back and the lifetime expectation score. Interpretation of relationships between factors and reasons for disposal, however, is difficult because all exterior traits were more or less included ina factor . Several exterior traitsjudge da tgil tstag e showed arelationshi p with reason for disposal. This is an indication that those traits have a predictive value for the gilt's chance toproduc e more than four litters. Results, however, could be biased by the prejudice of farmers and of inspectors,althoug hexterio rscore swer eno tsupplie d toth efarmers .Thi s biaswil l decrease with increasing time interval betweenmomen t ofjudge ­ ment and culling. Analysis of gilt's exterior, in a subset of data where allanimal sha dproduce d at leaston e litter, showed onlymino r changes in contrastestimates .Thi sbia s istherefor eassume d tob eabsent .Th e larger gilts (longeran dbroader )ar emor esusceptibl e tole gproblems .Thi smigh t be due to theirhighe r weight,makin ghighe r demands upon the strength of the legs. Predisposition to bowed legs in gilts is unfavourable. A slow gaitan dmuc h swinging of thebac kha s anundesirabl e correlationwit h leg weakness. The last two traits and lifetime expectation score are closely correlated. Those traits might be helpful to reduce early involuntary culling. Combination of those three traits might yield to an acceptable indicator for risk of culling because of legweakness .

75 ACKNOWLEDGEMENT

Parto fth ecost so fthi sstud ywer ecovere db yfinancia lsuppor tfro m theDutc hCommodit yBoar dfo rLivestoc kan dMeat .Th eessentia lcontribu ­ tion of the Dutch Pig Herdbook, thebreedin g organizations Cofokb.v. , Euribridb.v .an dNieu wDallan db.v .i nobtainin gth eexterio rjudgement s was greatly appreciated. Hendrix'Voeder sb.v .wa shelpfu l ingatherin g sow productivity data. We gratefully acknowledge the contribution and co-operationo f theherdsme nwhos eso wherd shav ebee njudge ddurin gtw o years.

REFERENCES

Bliss, C.I., 1970. Statistics in biology (Vol. II).McGraw-Hil l Book company,639pp . Bowden,V. , 1982.Typ e classification in dairy cattle:A review.Anim . Breed.Abstr .50 :147-162 . Grendalen,T. ,1974 .Osteochondrosi san darthrosi si npigs .VII .Relation ­ shipt ojoin tshap ean dexterio rconformation .Act aVet .Scand .46 :1-3 2 (suppl). Harvey,W.R. ,1977 .User' sguid efo rLSML7 6(Mixe dMode lLeast-Square san d MaximumLikelihoo dCompute rProgram) .Mimeo. ,Ohi oStat eUniv . Hetzer,H.O .an dMiller ,R.H. ,1972 .Correlate dresponse so fvariou sbod y measurements inswin eselecte dfo rhig han dlo wfatness .J .Anim .Sei . 35:743-751 . I.M.S.L., 1984.I.M.S.L . library,referenc emanual ,editio n9.2 ,Interna ­ tionalMathematica lan dStatistica lLibraries ,Houston ,Texas ,U.S.A . Pattison,H.D. ,1980 .Pattern so fso wculling .Pi gNew san dInfo .1 :215 - 218. Rogers, G.W. and McDaniel, B.T.,1989 .Th eusefulnes s of selection for yieldan dfunctiona ltyp etraits .J .Dair ySei. ,72 :187-193 . S.A.S.Institut e Inc.,1985 .User' sGuide :Statistics ,Versio n5 Edition . Cary,N.C. ,95 6pp . Vande rWal ,P.G. ,Goedegebuure ,S.A. ,Va nde rValk ,P.C. ,Engel ,B .an d VanEssen ,G. , 1987.Le gweaknes s andosteochondrosi s inpigs ;diffe ­ rencesbetwee nth esexe so ffou rbreeds .Livest .Prod .Sei .16 :65-74 .

76 VanSteenbergen ,E.J. ,1989 .Descriptio nan devaluatio no fa linea rscorin g systemfo rexterio rtrait si npigs .Livest .Prod .Sei .23 :163-181 . VanSteenbergen ,E.J. ,Kanis ,E .an dVa nde rSteen ,H.A.M. ,1990 .Geneti c parameterso ffattenin gperformanc ean dexterio rtrait so fboar steste d incentra lstations .Livest .Prod .Sei .24 :67-85 . Walstra,P. ,1980 .Growt han dcarcas scompositio nfro mbirt h tomaturit y inrelatio nt ofeedin gleve lan dse xi nDutc hLandrac epigs .Communica ­ tionsAgricultura lUniversit y80-4 ,Wageningen .

APPENDIX

Descriptiono f2 0exterio rtraits . Score Exteriortrait s 0 44 9 1.Connectiono fshoulde rblad ewit h body loose solid Back and loin 2.Length short long 3.Width narrow broad 4.Regularity irregular regular Hams 5.Length short long 6.Width narrow broad Forelegs 7.Frontvie wleg s 0-shaped X-shaped 8.Sidevie wleg s sickledstraigh t buckled 9.Sidevie wpaster n steepangl e lowangl e Rear legs 10.Rearvie wleg s 0-shaped X-shaped 11.Sidevie wleg s straight bowle g 12.Sidevie whoc kjoin t steep sicklehocke d 13.Sidevie wpaster n steepangl e lowangl e 14.Quantitysubcut .liqui d at hock joint much nil 15.Knottinesso flegs ;surfac e of tubercles > 12cm2 6 2 0cm 2 Claws rear leg cm 16.Ratiosiz einne ran doute r claw < h 3/4 >1 17.Toesiz e small large

18.Gaitpattern ;movement s slow(stiff ) easy,quic k 19.Swingingo fbac k nil much 20.Lifetimeexpectatio n short long

77 Chapter 5

EXTERIOR IN SOWS: 2.RELATIONSHIP S WITH REPRODUCTIVE PERFORMANCE AND LONGEVITY

E.J. van Steenbergen, E.Kanis ,W.J . Koops andH.A.M .va n der Steen

Department ofAnima l Breeding, Wageningen Agricultural University, P.O. Box 338,670 0A HWageningen ,Th eNetherlands .

Submitted for publication inLivestoc k Production Science. Reproduced by permission of Elsevier Science Publishers B.V., Amsterdam.

79 ABSTRACT

Over a two-year period, exterior of over 5000 sows on 24 commercial multiplierherd swa sappraised .Twent yexterio r traitswer ejudge dafte r eachweanin g andwer e scored linearly.Relationship s of exterior traits withreproductiv eperformanc ewer einvestigate dan dfoun dt ob eo fmargina l importance. For 2809 animals, of which most were purchased at 6 months of age, survivalanalysi swa suse dt oinvestigat eth erelationshi pbetwee nexterio r traits judged at gilt stage and longevity. A Weibull model was found adequately in describing the survival distribution. For most traits concerning the posture of the legs, both low and high scores were negativelycorrelate dwit hlongevity . Theeffec to fa chang eo fon estandar ddeviatio nuni t (s.d.)fro mth e averageexterio r scoreo nth emedia nlengt ho fproductiv e life (LPL)wa s estimated.Broade rhindquarter swer eunfavourabl yrelate dt oLP L(-5 3 days pers.d. )whil ewit hon es.d .smalle rhindquarter sLP Lwa sincrease db y4 4 days. LPL increased with decreased swelling at the hock joint. A more flexiblegait ,les sswingin go fth ebac kan da highe rlifetim eexpectatio n score coincided with a higher median LPL (67,4 4 and 42 days per s.d. respectively).Selectio no nexterio rtrait sjudge da tgil tstage ,wit hmai n attentionfo rgait ,ca nincreas eaverag elengt ho fproductiv e lifean ds o decreaseth eannua lcullin grate .

INTRODUCTION

Exterioro fanimal sha sbee nan dstil li sa topi cwhic hdraw sattentio n of farmers,breeder s and scientists. Main attention is focused on the predictivevalu eo fexterio r traits forinvoluntar y culling (e.g.i nth e case ofhealt h and fertility problems and leg weakness). A decrease of involuntarycullin gincrease sselectio npossibilitie san dmigh tals oresul t ina smaller portion of first litter sows,whic h results ina positive effect onnumbe r ofpig spe r sowpe ryear .Economi c impact ofprematur e disposal was calculated by Dijkhuizen et al. (1989) and turned out to average Dfl. 124 per culled sow and Dfl. 183 when the sow was culled becauseo fle gweakness/lameness .

81 Sow'sexterio rma yb euse dt opredic the rlongevit yan dfutur ereproduc - tivity.Thi spredictio n influences thedecisio nwhethe ro rno t tocul l the sow.Th e relevance of exterior intha tdecisio ndepend so navailabilit y of additional information especially concerning the animals reproductive ability. For gilts little additional information is available and future expectation of longevity and reproductivity might largely be based on the animalsexterior .Economi c relevanceo fexterio rtrait sca nb ederive d from the relationships of those traitswit h longevity and reproductivity. Reviews of studies concerning relationships of exterior traits with production and longevity in dairy cattle are given by Bowden (1982) and Burnside et al. (1984). The latter authors concluded that there is still no definite answer on whether or not exterior judgements are useful to predict longevity. They concluded that the relative economic weights for sirebreedin g values of milk production per lactation and stayability (as simple indicator of longevity) are about 2:1. Rogers and McDaniel (1989) concluded that production in first lactation should receive about two to three times as much emphasis as the combination of three exterior traits thatwer e associated with involuntary culling. Drobig et al. (1983) and Schlegel et al. (1983) studied relationships ofbod y measurements of gilts with reproductive performance. They showed phenotypicassociation sbetwee nbod ysiz eparameter san dreproduction .Kin g (1989) concluded that reproductive efficiency of gilts after parturition wasno t significantly affectedb ybod yweigh to rbod y composition atabou t 170day so fage .Rothschil d etal . (1988)reporte dcorrelate d responses in reproduction after 5 generations of divergent selection for front-leg weakness in Duroc pigs. Although no obvious trends could be observed, results suggested, in general, that culling for front-leg weakness would make some improvement in reproduction productivity. In a previous study of linearly scored exterior traits of gilts, Van Steenbergen et al. (1990) studied associations between exterior traits judged at gilt stage and culling reasons.Significan t differences existed fora numbe r of traitsbetwee n the group ofanimal s thatproduce d at least four litters and the group of animals that were disposed because of leg weakness before the 4th parity. In the present study, associations of exterior of gilts and sows with reproductive performance and of exterior traits judged at gilt stage with longevity are analyzed.

82 MATERIAL

In198 5an d1986 ,gilt san dsow sfro mdifferen tstrain swer ejudge dfo r exterior at 24commercia lmultiplie r herds of size 100-200 sows.Eleve n inspectors ofth eDutc hherdboo kan do fthre ebreedin gcompanie srotate d over the farms to score exterior of the animals. In total 20exterio r traits (VanSteenbergen ,1989 )wer e scoredo na linea rscale .Farm swer e monthlyvisite db ya ninspecto rwh ojudge dal lgilt so fapproximatel y 7-9 monthso fag e (giltstage )an dal lsow sfro mweanin g (onaverag e2 8day s afterfarrowing )til l4 0day spos tweaning .Mos tgilt swer ejudge dtwic e by different inspectors and most sows once after each weaning of the piglets. Data enclosed 5564 animals,bor n from 1979 to 1985, of which exterior scoresan dreproductiv eperformanc e ofa tleas ton elitte rwer e available. Over 50Z (2812)o f the animals had been judged at the gilt stage.Reproductio nproductivit ydat awer eobtaine d in1988 .A ttha ttim e over70 %o fal lanimal swer eculle dfo rvariou sreasons .A mor edetaile d description of the material is givenb yVa n Steenbergen et al. (1990). Informationo nreproductio ntrait swer epe rlitter : -tota lnumbe ro fpiglet sbor n (TNB), -numbe ro fstill-bor npiglet s (NSB), -pigle tmortalit yi nth esucklin gperio d(PMS )(numbe ro fpiglets )an d -interva lweanlng-oestru s(IWO )(numbe ro fdays );th einterval sweaning - oestrusconcerne dth eperiod safte rfirs tt onint hlitter . Threemeasure sfo rso wlongevit ywer eused : -lengt h ofproductiv e life (LPL)a s difference between date offirs t servicean ddat eo fdisposa l (numbero fdays) , -numbe ro flitter sproduce d (NLP)an d -tota lnumbe ro fpiglet sproduce d (NPP).

METHOD

Exterioro feac hanima lwa scharacterize db yth eexterio rscore sa sgil t (GE)and ,a tolde r agesb y theaverag e deviation of theexterio r scores fromth eparit y leastsquare smean s (DM)an dth edeviatio no fth elinea r regressiono fth eexterio rscore so nparit ynumber ,fro mth eaverag elinea r regression (DLR) was estimated. This was done for all 20 traits. The

83 original exterior scores were transformed by a logit transformation which changed therang eo fscore so f0 t o9 int o -4.5t o4.5 .Principa l component analysis was used to reduce the 60 exterior variables (20 traits forGE , 20 traits for DM and 20 traits for DLR) to a more feasible number of 15 factors (Van Steenbergen et al., 1990). An effect of culling on parity estimates of reproduction traits is likely to exist but is not accounted for with ordinary Least Squares analyses.Reproductio ntrait swer eanalyze db yth eLSML7 6progra mo fHarve y (1977)wit h themodel : yijton -/ *+ HSi+ A u + YSk + Pm+ eljknm (1) where, yijkran "a reproduction trait li - mean

ESt - fixed effect ofherd-strai n group i (i-l,...,74)

Atj = random effect of animalj withi nherd-strai ni

YSk - fixed effect ofyear-seaso n k (k-l,...,16)

Pm - fixed effect ofparit y m (m=»l 10) e_ijkmn - random error term

Seasons were based on month of conception and included the periods April to September and October to March. For TNB the number of matings per oestrus (1o r 2)whic h resulted inpregnanc y was included in themode l as an extra factor. Reproductive performance of each animalwa s characterized ina similar way aswa s done for exterior (Van Steenbergen et al., 1990). For each of thefou rreproductio n traits theaverag e ofth edeviation s from theparit y

leastsquare smean s (TNBj^ NSBj,PMS Xan dIUO ^ andth elinea rregressio no f

thatdeviation s onparit ynumbe r (TNB2,NSB 2, PMS2an d IW02)wer e estimated per animal using orthogonal polynomials. Positive values for TNB2, i.e. a more than average increase ofnumbe r ofpiglet sborn ,wit h parity number,

andnegativ evalue s forNSB 2, PMS2, and IW02ca nb econsidere d asdesirabl e andindicat etha tth efunctionalit yo fbiologica lprocesse swhic hdetermin e reproductive performance, increases with parity number. The association between the eight reproduction variables and the exterior variables was based on all 5564 animals and was analyzed by the GLM procedure of SAS (SAS Institute Inc., 1985). The model to analyze

84 variance in the reproductionvariable s includedherd-strai n group and all 15 exterior factors as covariables (model 2).Relationship s of exterior traits judged as gilts with reproduction were studied with stepwise regression. Reproduction variables were analyzed with a regression model which included scores and squared scores of exterior traits and the herd-strainvariabl e (model 3). Thedefinit e regressionmode lwa s searched by stepwise regression performed by the STEPWISE procedure (method-step- wise) of SAS (SAS Institute Inc., 1985). The herd-strain variable was forced into the model while the criterion for exterior variables to be addedt oo rdelete dfro mth emode lwa sbase do na F statisti cleve lo f 0.15.

Survival analysis.

At themomen t that sow reproduction productivity databecam e available foranalysis ,abou t 30%o fth eanimal swit hexterio rjudgement swer e still in production. Complete longevity measures could only be computed for animals thatwer e culled. Information on animals still alive,however , is useful. The statistical technique of survival analysis utilizes both the completean d incomplete (censored)observation s (Kalbfleischan dPrentice , 1980). A first step in survival analysis is the estimation of the distributiono fth elongevit ymeasures .Whe nthi sdistributio ni sspecifie d by a function, the following model can be assumed for analyzing the longevity measures:

Y •=I + SfliX i+ SajZ j+ aW (4)

where Y is the natural logarithm of a longevity measure, I is the inter­

cept, ßjar e unknown coefficients of covariates Xlt a, are the constants of the j levels of factor Z, a is a scale parameter and W is avecto r of errors from the assumed distribution. The covariables in the model were: i) average number of piglets born, calculated as deviate from the herd- strain mean, and ii) one exterior trait (linear and quadratic). Herd was considered as avariable . Ducrocq (1987) showed the Weibull function adequately describing the distribution of length of productive life in dairy cattle. The Weibull function [S(t)- exp[-(At)p]where ,S(t )wa s the fraction ofanimal s still alivea tt ,wit h tbein g alongevit ymeasure ,an dA , pparameters ]use d for

85 thedescriptio no fth esurviva ldistribution ,wa sempiricall yvalidate db y plotting ln(-ln[S(t)]) versus ln(t). If the assumed distribution is correct, then the plot should give approximately a straight line. Parameters ofth eWeibul lmode lca nb ederive dfro mtha tline ;th eslop e providesa nestimat eo f pan dth eintercep ta nestimat eo f pln(X). Theeffec to fexterio rtrait so nlongevit ymeasure swa squantifie db y estimationo fth emedia no fth elongevit ymeasure sfo rth eaverag eexterio r scores(AES )an dfo rth eAE S± on euni tstandar ddeviatio n(s.d.) .Surviva l analyseswer eperforme dwit hth eLIFERE Gprocedur e ofSA S (SASInstitut e Inc.,1985) .

RESULTS

Leastsquare smean s (model1 )o fth ereproductio n traits forparit y1 to1 0ar e graphically presented inFigur e 1.A n increase of 2.5 piglets born can be noticed from first to sixth parity. Number of still-born piglets increases from 0.5 in parity 1 to 2.3 in parity 10. Piglet mortality inth esucklin gperio d increasesunti lsevent hparit y from 1.2 to over 3 piglets per litter. Interval weaning-oestrus shows a rapid declineafte rfirs tparit yan dstabilize safte rthir dparit ya ta leve lo f approximately 7days .A secon dmatin gdurin gth esam eoestru s resultsi n 0.2extr apiglet sbor n(P<0.001) . Exterior factors whichhav e a significant effect on the reproduction variables (model2 )ar esummarize di nTabl e1 .Gilt' sexterior ,especiall y therea r legs (traits 14-20; seeAppendices) ,an d sow's exterior traits sidevie wrea rleg s (11),gai t(18) ,swingin go fth ebac k(19 )an dlifetim e expectation (20) are associated with most reproduction variables. Regression coefficients of the regression of reproduction variables on exteriortrait sjudge da tgil tstage ,tha tremaine d inth efina lstepwis e regression model (model 3),ar e in Table 2.Averag e 'total number of pigletsborn ' (TNBX)increase swit h lengtho fth egil t (2)an ddecrease s withhighe rscore sfo rregularit yo fbac kan dloi n(4) .Bot haverag enumbe r of still-bornpiglet s (NSBj)an d the regressiono fNS B onparit y number

(NSB2)sho wa negativ erelatio nwit hth elinea ran dquadrati ccomponen to f sidevie wrea r legs (11).Referrin gt oth enumbe ro fstill-bor npiglets , bowleg sar efavourable .Pigle tmortalit ydurin gth esucklin gperio d(PMS ^

86 Average number of piglets bom Average mnbai of still-born piglets Piglets 44

H i i i 123456789 10 Parity number

Piglet mortality during suckling period Intervalweaning-oestru s Days 15

14

13

12

11

10

S

6

7

e

5 4 5 6 7 12 3 4 5 6 7 8 9 10 Parity number Parity number

Figure1 .Leas t squaresmean s (+S.E. )o freproductio n traitsb yparit y number.

87 isinfluence db y thequadrati c componento fknottines s (15),an db yth e linearcomponen to fcla wrati o (16),an dgai t (18).Th eexterio rcharac ­ teristicsa tgil tstage ,eve nsize dclaw san da slo wgai tar eunfavourabl y correlatedwit hpigle tmortalit ydurin gth esucklin gperiod .

Table1 .Factor swit ha significan teffec to nreproductio nparameters .

Reprod. param1. Factornumber 2an dsignificanc eleve l TNBj •I* r\* Q***

TNB2 5* NSBi 1* * f.*** .* Ç**

NSB2

PMSX 1 ,2 ,4 ,5 , 7 ,1 0 ,13

PMS2 1",4*** ,5* .7 " IWOi 1*.2* *

IW02 2**,10 "

TNB -Tota lnumbe ro fpiglet sbor n NSB -Numbe ro fstill-bor npiglet s PMS- Pigle tmortalit y inth esucklin gperio d IWO- Interva lweaning-oestru s Subscript1 : deviatio nfro mth emea n(DM ) Subscript2 :regressio no fD Mo nparit ynumbe r SeeAppendi x2 .

Survivalanalysis .

Agraphica lvalidatio no fth eWeibul lmode lfo rLPL ,NL Pan dNPP ,base d on566 0sows ,i si nFigur e2 .Th eregressio no festimate so fln(-ln[S(t)] ) on ln(t) showed that estimates of S(t) can be well approximated by a Weibull function:Coefficient so fdeterminatio n (R2)fo rLPL ,NL Pan dNP P ofove r0.9 7wer eobtained . Regressioncoefficien t (r.c.)estimate s (model4 )fo rlnY ,wher eY i s alongevit ymeasure ,o nexterio rtrait s(linea ran dquadratic )ar ei nTabl e 3.Whe n |r.c.|i ssmal l (e.g.<0.20) , thenr.c .ca nb e interpreted asa proportion;i.e .a r.c .o f-0.1 0stand sfo r10 %decreas eo fY .A linea r

88 Table 2.Regressio ncoefficient s of reproductionvariable s on gilt's exterior traits that were in the final stepwise regression model.Underline d coefficients concernsquare dexterio rscores .

Reproductionvariable s

Gilt'sexterio rtrait s TNB, TNB, NSB, NSB, PMS, PMS, IWOj IUO,

1.Shoulde rblad e .05 -.06 Back and loin 2.Lengt h .38*" .09 -.24 3.Widt h .09 4.Regularit y -.25" -.14" 0.30 Hams 5.Lengt h .10 6.Widt h -.07 -.04 Forelegs 7.Fron tvie wleg s -.14 8.Sid evie wleg s .16* 9.Sid evie wpaster n -.15 .10* .30 Rear legs 10.Rea rvie wleg s -.26 -.26* .60" 11.Sid evie wleg s -.25"* -.09 .49 .34 AS"' .15 .48* 12.Sid evie whoc kjoin t -.12 -.13 14.Liqui da thoc kjoin t .09 -.05 15.Knottines so fleg s -.07 .10"* Claws rear leg 16.Rati osiz eo fclaw s 04 .12* .09 17.To esiz e 08** 18.Gait ;movement s -.07 -.10* -,07 20.Lifetim eexpectatio n .12"

effect of exterior as gilt on ln(LPL) is shown for connection of shoulder blade (1),sid e view hock joint (12),swellin g at the hock joint (14), knottiness of the rear legs (15),gai t (18),swingin g of thebac k (19)an d lifetime expectation score (20).Linea r regression coefficients (a) of ln(NLP) and ln(NPP) on the three last mentioned exterior traits are also

89 li.[-ta]"11.33+1.620.1n(LFD R -0.9932 ln[-ln(S(t))]—2.77+1.481'ln(NLP) R -0.9762

ln[-ln(S(t))) 2

ln(LPL) ln(NLP)

ln(-ln(S(t))l -6.58+ 1.512'ln(NPP) R-0.98« « ln[-ln(S(t))] 2

Fipure2 .Chec ko fth eWeibul lmode lassumptio nfo rth esurviva ldistribu ­ tion of length of productive life (LPL), number of litters produced(NLP )an dnumbe ro fpiglet sproduce d(NPP )b yregressio n ofln[-ln(S(t)) ]o nln(LPL) ,ln(NLP )an d ln(NPP).

highlysignificant .A negativ eassociatio no flongevit ymeasure swit hth e quadraticcomponen to fwidt ho fbac kan dloi n(3) ,forele gstatur e(7,8,9) , sidevie wrea rleg s(11 )an dhoc kjoin t(12 )i sfound ,whic hindicat etha t lowan dhig hscore sfo rthos etrait sar eunfavourabl erelate dt olongevity . Effects of exterior on NLP and NPP are similar to LPL, which is not surprising because of thehig h correlation (>0.90 )betwee n those three longevitymeasures .Furthe r resultswil l thereforeonl yb epresente dfo r LPL. Formos texterio rtrait sa negativ eregressio ncoefficien tb i sfoun d whichpoint st oreduce dpossibilitie so fselectio no nthos eexterio rtrait s to increase LPL. Effect of exterior traits on LPL is demonstrated in Figure3 .Calculatio ni sbase do nregressio ncoefficient sa an db i nTabl e3 .

90 Table3 . Regressioncoefficien testimate san dsignificanc elevel so flifetim e measures on gilt's exterior traits (linear (a)an d quadratic(b)) .

Lengthpro d life Number of litters Number of piglets

Gilt's exterior traits a b a b a b

1. Shoulder blade .122" - .063 .084* -.041 .097* -.048 Back and loin 2. Length -.001 .073 -.040 -.056 -.042 -.064 3.Widt h -.031 .109* * -.009 -.096" -.010 -.109** * 4. Regularity .067 .057 .034 -.038 .043 -.045 Hams 5. Length -.043 .001 -.061 .018 -.072 .021 6. Width -.094 .029 -.076 -.025 -.088 -.027 Forelegs 7. Front view legs .046 .078* .024 -.097*** .026 -.104** * 8. Side view legs -.064 .077** -.057 -.057* -.063 -.065* 9. Side view pastern -.007 .075** -.018 -.074** -.020 -.081** Rear legs 10. Rear view legs .024 .079 -.004 -.010 -.007 -.010 11. Sidevie w legs -.017 .088*** -.005 -.060" -.014 -.065" 12. Side viewhoc k joint .096* - .153** .075 -.100 * .075 -.118* 13. Side view pastern .043 .066 .042 -.042 .040 -.047 14. Liquid athoc k joint .071* .006 .037 .008 .045 .008 15. Knottiness of legs .064* - .012 .051 -.006 .055 -.006 Claws rear leg 16. Ratio size of claws -.041 .002 -.072 .024 -.082 .028 17. Toe size .019 .037 -.038 .051 -.044 .058

18. Gait; movements .179***- .031 .142** * -.019 .163*** -.023 19. Swinging of back - 143***- .047 -.094" -.028 -.109*" -.032 20. Lifetime expectation .065***- .010 .055*** -.001 .062*** -.002

A low score forwidt h of theham s results ina highe r median LPL (44day s per s.d.) while a higher score has an even stronger opposite effect (-53 days per s.d.). Both a lower and higher than average score for side view pastern forelegs result in a small decrease of the median LPL (12 and 9 days respectively). It canb e seen that a change of the average scores of exterior traits 7 to 13 (fore- and rear legs) mostly results in a considerable decrease of the median LPLwhil e only marginal increases can be noticed. When the average exterior score changes one unit s.d., an increase (>3 0days )o fth emedia nLP Li snoticed :4 4day s for smallerham s

91 675 700 725 750 775 800 825 850 (Days) Exterior craits

1.Shoulde rblade . Back and loin 2.Lengt h

3.Widt h

4. Regularity. Hams 5.Lengt h

6.Widt h Forelees 7.Fron tvie w legs.

i.Sid evie w legs.

9. Sidevie wpastern.... - Rear lees 10. Rear view legs

11. Sidevie w legs

12. Sidevie whoc kjoint .

13. Sidevie wpastern... .

14. Liquid athoc kjoint .

15.Knottines s of legs... Claws rear lee 16.Rati o sizeo fclaws.. -

17. To e size

18.Gai t;movement s - < 19. Swinging ofbac k

20. Lifetime expectation.

Fipure 3. Estimates of themedia no flengt h ofproductiv e life for average exterior scores (•) and the average plus or minus one unit s.d. denoted by t>an d< , respectively .

92 (6), 36day sfo rles sswellin ga tth ehoc kjoin t (14),6 7day sfo reasie r movements/gait (18), 4 4 days for lessswingin go f thebac k (19), an d4 2 daysfo ra highe rlifetim eexpectatio nscor e(20) .

DISCUSSION

Parityeffec to fth ereproductio ntrait sa sdemonstrate di nFigur e1 i s biasedbecaus e of culling. Thisbia s increaseswit hparit y number while theseverit yo fth ebia sdepend so nth ecorrelatio no freproductivit ywit h theris kt ob eculled .Fo rtota lnumbe ro fpiglet sbor ni ti slikel ytha t differencesbetwee nfirs tan dhighe rparit ynumber sar eover-estimated . From Table 1 itwa s concluded that a linear combination of exterior traitsconcernin gth erea rleg sa tgil tstag e (factor1 ;trait s14-20 )i s associatedwit hreproductivity .Result so fth eanalysi so fG Etrait s(Tabl e 2)sho wonl ymino rassociation so fgilt srea rleg swit hreproductivity ,an d therefore do not confirm the conclusion based on Table 1. A possible explanation for this discrepancy mightb e that results of Table 1wer e basedo nal l (5564)animals ,whil eanalysi so fexterio r traitsjudge da t giltstag ewer ebase donl yo nanimal stha tha dbee njudge dwhe ngil t(281 2 animals).Fro mTabl e2 i tca nb econclude dtha texterio rtrait sjudge da t gilt stage and reproductivity have just small associations. This is in accordancewit hresult so fDrobi ge tal .(1983 )an dSchlege le tal .(1983) . Thedistributio no fth elongevit ymeasure scoul dadequatel yb edescribe d by theWeibul lmodel ,whic h isi nagreemen twit h findings for lengtho f productivelif eo fdair ycattl e(Ducrocq ,1987) .Cullin grate sfo rsow spe r parity areofte nrequire d inmode lcalculation s (e.g.Dijkhuize ne tal. , 1986;D eVries , 1989). Those culling rates (CR)ca nb e derived from the survival distribution as: -dln(S(t))/dt . For theWeibul l function this leads to CR(t) - pAp(t)''~1. Parameters of the Weibull function can be derivedfro mth eequatio nln[-ln(S(t)) ]- pln(A )+ pln(t )an dwere :p=1.48 1 andA-0.154 1 (Fig. 2).Ris k fora so w tob e disposed afterparit y tca n thenb ecalculate da sCR(t)=0.09t 0*81. MedianLP L (Fig. 3)average d 770days ,s o50 %o fth esow s reached an ageo fa tleas t100 0day swhic hstand sfo ra productio no f4. 8 litterso n average.Thi si salmos ton elitte rabov eth efigur epresente db yDijkhuize n etal . (1989),wh o found 51%o fth esow s toproduc ea t least4 litters .

93 Thatlowe rfigur emigh tb edu e toth elarg eportio no fremove dfirs tparit y sows in the data of lastmentione d authors. Inmos t of the 24 commercial multiplierherds ,gilt s were purchased at 6 months age. A considerable selection on exterior might have occurred beforedelivery . Inspit eo fth ereduce dvariatio ni nexterior , significant associations of exterior with longevity measures are observed. The following exterior traits,judge d at gilt stage,hav e an important effect onlongevity :widt h of theham s (6),swellin g atth ehoc kjoin t (15),gai t (18), swinging of the back (19) and lifetime expectation score (20).O f those five traits, lifetime expectation (20)wa s scored most subjectively because it implied an overall impression. This trait might therefore be less attractive for selection purposes. Longevity shows a negative relationship withbroa d size andwit hextrem e scores for sidevie w ofmos t leg traits.Littl e swelling at thehoc k joint,a quic k and flexible gait, and little twisting of the hind quarters are favourably correlated with longevity.Thos e five traitswer eals o indicatedb yVa nSteenberge n etal . (1989) to be relevant in relation to disposal for leg weakness. It is therefore likely that increased longevity through selection on exterior traits, will be achieved mainly by a reduced culling because of leg weakness. An increasedher d lifewa s found tob eeconomicall ybeneficia lb y Kroes and Van Male (1979) and Dijkhuizen et al. (1986). The latter author calculated 10 litters to be the economic optimum herd life for average producing sows. Dijkhuizen et al. (1989) calculated economic losses for sows culledbecaus e of legweakness ,averagin g Dfl. 183pe r culled sow.A strong selection in gilts on the three closely correlated exterior traits gait (18),twistin g of hind quarters (19) and lifetime expectation (20) will reduce involuntary culling because of leg weakness and increases median length of productive life by approximately 75 days which is equivalent to 0.5 parity. Selection on exterior might therefore expected toyiel d substantial economicbenefits .

ACKNOWLEDGEMENT

Part of the costs of this study were coveredb y financial support from the Dutch Commodity Board for Livestock andMeat .Th e essential contribu-

94 tion of the Dutch Pig Herdbook, the breeding organizations Cofok b.v., Euribridb.v .an dNieu wDallan db.v .i nobtainin gth eexterio rjudgement s was greatly appreciated. Hendrix'Voeder sb.v .wa s helpful ingatherin g sowproductivit ydata .W egratefull yacknowledg eth econtributio nan dco ­ operationo fth eherdsme nwhos eso wherd shav ebee njudge ddurin gtw oyears .

REFERENCES

Bowden,V. , 1982.Typ e classification in dairy cattle:A review.Anim . Breed.Abstr .50 :147-162 . Burnside,E.B. ,McClintock ,A.E .an dHammond ,K ., 1984 .Type ,productio n andlongevit yi ndair ycattle :A review .Anim .Breed .Abstr .52 :711-719 . DeVries ,A.G. ,1989 .A mode lt oestimat eeconomi cvalue so ftrait si npi g breeding.Livest .Prod .Sei .21 :49-66 . Dijkhuizen,A.A. ,Morris ,R.S .an dMorrow ,M. ,1986 .Economi coptimizatio n ofcullin gstrategie si nswin ebreedin gherds ,usin gth e"PorkCHO Pcompute r program".Prev .Vet .Medic ,4 :341-353 . Dijkhuizen, A.A., Krabbenborg, R.M.M, and Huirne, R.B.M., 1989. Sow replacement: a comparison of farmers' actual decisions and model recommendations.Livest .Prod .Sei .23 :207-218 . Drobig,M. ,Ritter ,E .an dRitze ,W. ,1983 .Untersuchunge nübe rZusammen ­ hängezwische nKörpermasse nsowi eAnsatzleistun gun dde rWurfleistun gbe i Schweinenunterschiedliche r genetischer Konstruktion.Arch .Tierz .26 : 173-185. Ducrocq, V.P., 1987.A n analysis of length ofproductiv e life indair y cattle.Thesi sCornel lUniversity ,21 6pp . Harvey,W.R. ,1977 .User' sguid efo rLSML7 6 (MixedMode lLeast-Square san d MaximumLikelihoo dCompute rProgram) .Mimeo. ,Ohi oStat eUniv . Kalbfleisch, J.D. and Prentice,R.L. , 1980.Th e statistical analysis of failuretim edata .Joh nWile yan dSons ,Ne wYork ,NY . King,R.H. ,1989 .Effec to fliv eweigh tan dbod ycompositio no fgilt sa t 24week so fag eo n subsequentreproductiv e efficiency.Anim .Prod .49 : 109-115. Kroes,Y .an dVa nMale ,J.P .,1979 .Reproductiv elifetim eo fsow si nrelatio n toeconom yo fproduction .Livest .Prod .Sei. ,6 :179-183 .

95 Rogers, G.W., and McDaniel, B.T., 1989. The usefulness of selection for yield and functional type traits.J . Dairy Sei. 72:187-193 . Rothschild,M.F. ,Christian ,L.L .an dJung ,Y.C. ,1988 .Geneti c control of front leg weakness in Duroc Swine. II. Correlated response in growth rate, backfat, and reproduction from five generations of divergent selection. Llvest. Prod. Sei. 19:473-485 . S.A.S. Institute Inc.,1985 .User' s Guide: Statistics,Versio n 5Edition . Cary,N.C. , 956pp . Schlegel, W., Wähner, M. and Wiese, H.P., 1983. Untersuchungen zur Abhängigkeit der Fruchtbarkeitleistungen bei Jungsauen von Ihren Körpermassenbe i Zuchtbenutzungsbegin. Arch. Tierz. 26:341-349 . VanSteenbergen ,E.J .,1989 .Descriptio nan devaluatio no fa linea rscorin g system for exterior traits inpigs .Livest .Prod . Sei. 23:163-181 . Van Steenbergen, E.J., Kanis, E.,Koops ,W.J . and Van der Steen,H.A.M. , 1990. Exterior insows : 1. Effect ofparit y number and association with reasons for disposal. Submitted toLivest . Prod. Sei.

96 APPENDICES

Appendix1 .Descriptio no f2 0exterio rtrait s Score Exteriortrait s 0 4H 1.Connectiono fshoulde rblad ewit hbod y loose solid Back and loin 2.Length short long 3.Width narrow broad 4.Regularity irregular regular Hams 5.Length short long 6.Width narrow broad Forelegs 0-shaped X-shaped 7.Frontvie wleg s sickledstraigh t buckled 8.Sidevie wleg s steepangl e lowangl e 9.Sidevie wpaster n Rear legs 0-shaped X-shaped 10.Rearvie wleg s straight bowle g 11.Sidevie wleg s steep sicklehocke d 12.Sidevie whoc kjoin t steepangl e lowangl e 13.Sidevie wpaster n much nil 14.Quantitysubcut .liqui da thoc kjoin t > 12cm2 6cm2 0cm 2 15.Knottinesso flegs ;surfac eo ftubercle s Claws rear leg < H >1 16.Ratiosiz einne ran doute rcla w small large 17.Toesiz e slow(stiff ) easy,quic k 18.Gaitpattern ;movement s nil much 19.Swingingo fbac k short long 20.Constitution

Appendix2 • Traitswit ha nabsolut eloadin go nth efactor sgreate rtha n0.5 0

Exteriortraits 1 Factor Exterioro fgilt s Leveldeviatio n Coursedeviatio n 1 14t o2 0 2 11,1 8t o2 0 3 18'to2 0 4 3an d6 5 1,2 ,4 an d5 6 i,2, 4 and'5' 7 9,1 1t o1 3 8 ièan d1 7 9 3an d6 10 16an d1 7 11 1an d4 12 11t o1 3 13 11an d1 2 14 3an d6 15 9'and1 3 :SeeAppendi x1 fo rexplanatio no fth etrai tnumbers .

97 Chapter6

GENERALDISCUSSIO N

Therelevanc e of selectionfo rexterio r traitsdepend so nth epurpos e for which animals are domesticated. For small pets, fancy poultry and pigeonswhic h arebre d forexhibitio npurposes ,benefi t isdirectl y and mainlyrelate dt oexterio rtrait san dcharacter .Th emai npurpos eo fmos t livestockspecies ,however , isproductio no ffoo dfo rhuma nconsumption , e.g.milk ,mea tan deggs .Economi cbenefi to fthos especie sca ndirectl y be related to amount and quality of their products. The importance of exteriortrait sshoul dtherefor edepen do nth erelationshi po fthes etrait s witheconomi cmerit .Selectio no nexterio rtrait sshoul donl yb edon ewhe n shorto rlonge rter meconomi cmeri ti sexpected . In this thesis the relevance of exterior traits for a pig breeding programme has been studied. Selection on exterior traits can influence selection response for growth performance and reproduction traits, dependingo nheritabilitie san dgeneti can dphenotypi cinterrelationships . Nowadays attention ismainl y focused on those exterior traitswhic h are expectedt ob erelate dt oconstitution ,especiall ylocomotor yability .I n this chapter results of chapters 2 to 5ar e discussed invie w of their practical application.Mos trelevan tparameter s for thisdiscussio nare : repeatability,reproducibility ,standar ddeviation ,heritability ,correla ­ tions with growth performance and relationship with constitution and longevity.Thes eparameters ,whic har epresente di nforegoin gchapters ,ar e summarized inTabl e1 .

Scoringsystem .

Afirs tan dmai nste pwhe nconsiderin gexterio rtrait si st oestablis h a clear and conclusive definition of the characteristics one wants to describe and investigate. In dairy cattle, linear scoring systems for evaluation of exterior traits are considered to be favourable over descriptively scored traits (Thompsone t al., 1981 )an d arewidel y used now.Th enumbe r ofcategorie suse d toscor e thetraits ,however ,varie s overcountrie san drange sfro m9 t o6 0 (Diersan dSwalve ,1987 ;Thompso n

99 Table1 .Relevan tparameter sfo rth ediscussio no fth epractica lapplicatio n ofexterio rappraisa li npigs .

Measur Lng Relati anwit h Relation with accuracy performance longevi ty3 2 Exte r iortrait s *1 r2 s.d. h ADG UB a b 1. Shoulderblad e 0.61 0.38 1.30 0.15 0.24 0.51 0.12 -.06 Back and loin 2. Length 0.40 0.37 1.07 0.28 0.05 0.03 -.00 -.07 3. Width 0.54 0.35 1.05 0.38 0.41 0.26 -.03 -.11 4. Regularity 0.59 0.41 1.24 0.30 0.29 0.47 0.07 -.06 Hams 5. Length 0.40 0.12 0.96 0.06 0.41 0.05 -.04 .00 6. Width 0.74 0.62 1.29 0.36 0.34 0.05 -.09 -.03 Forelegs 7. Frontvie wleg s 0.35 0.20 0.77 0.06 0.13 0.10 0.05 -.08 8. Sidevie wleg s 0.49 0.38 1.11 0.06 0.14 0.08 -.06 -.08 9. Sidevie wpaster n 0.54 0.41 1.14 0.31 0.41 0.14 -.02 -.08 Rear legs 10. Rearvie wleg s 0.47 0.27 0.65 0.22 -.26 0.04 0.02 -.08 11. Sidevie wleg s 0.48 0.28 0.98 0.23 0.03 -.16 -.02 -.09 12. Sidevie whoc kjoin t 0.49 0.22 1.05 0.24 -.53 -.23 0.10 -.15 13. Sidevie wpaster n 0.40 0.28 1.04 0.30 -.18 0.06 0.04 -.07 14. Liquida thoc kjoin t 0.59 0.32 1.38 0.09 -.51 0.38 0.07 .01 15. Knottinesso fleg s 0.58 0.46 1.43 0.01 0.03 0.07 0.06 -.01 Claws rear leg 16. Ratiosiz eo fclaw s 0.55 0.40 1.36 0.09 -.06 -.25 -.04 .00 17. Toesiz e 0.45 0.30 1.27 0.15 -.06 -.24 0.02 .04

18. Gait;movement s 0.46 0.42 1.37 0.13 -.35 0.26 0.18 -.03 19. Swingingo fbac k 0.33 0.28 1.63 0.13 0.12 -.11 -.14 -.05 20. Nipples -- -- 0.08 ------21. Lifetimeexpectatio n 0.57 0.42 1.94 0.16 -.02 0.11 0.07 -.01

repeatability(r x)an dreproducibilit y(r 2)estimate s(Chapte r2 ;Tabl e5 ) genetic correlations for average daily gain (ADG) and ultrasonically measuredbackfa tthicknes s (UB).(Chapte r3 ;Tabl e6 ) Linear (a)an d quadratic (b) regression coefficient estimates of the naturallogarith mo flongevit y(ln(days) )o nexterio rtrait si nth egilt . (Chapter5 ;Tabl e3 ) etal ,1981 ;D eGraa fe tal ,1986 ;Murph y andSinnott ,1989) .Historica l factorsan dpersona lpreference swer eimportan ti nth echoic eo fth enumbe r ofcategorie s for linear scoring systems indair y cattle.N o conclusive rulesar eavailabl et odetermin e thenumbe ro fcategorie s tob eused .Th e rangeo fscores ,however ,shoul dsuppl ysufficien topportunit y toexpres s

100 thevisua l variation which can be detected. Both aver y small and a very large number of categories are supposed to reduce the reliability of the scores, and so negatively influence further use of the scores (e.g. for breeding value estimation). Inchapte r 2, 21exterio r traitswer e definedan ddescribe d ona linear scale from0 to9 wit h 0.5 intervals (19categories) . Inconsultatio nwit h inspectors who used the scoring system, and taking possible scoring forms into account, 19 categories were considered tob e a good choice. For all exterior traits large differences in mean and standard deviation between inspectors were found. These differences hinder application of linear scoring of exterior traits in pig breeding. An intensive and regular training of the classifiers in order to get uniform means (4.5) and standard deviations,i stherefor e necessary. Several inspectorsmad emino r use of the 'half' scores. Training or change to another scale might dissolve this preference for specific categories. Because the 'half' categories are proportionally distributed on the scale of 0 to9 ,dispro ­ portionateus eo fcategorie s forstatistica lanalysi so rfo rbreedin gvalu e estimation is expected tob e ofmino r importance.

Repeatabilities were used to determine the accuracy ofvisua l exterior appraisal.N ovisibl evariatio nwa sobserve d innippl e quality.Tha t trait was therefore excluded from further analysis.Fron tvie w forelegs and rear view rear legs showed little variation, while accuracy of appraisal was relatively small. These traits should therefore be considered of minor importance for selection purposes. A low accuracy for traits showing averagevariatio n (s.d. 1)migh tb e causedb y an incomplete and indistinct definition and description of the trait. This holds especially for length ofbac kan d loin,lengt ho fhams , andswingin go fth eback .A mor e distinct and comprehensive definition of these traits is therefore recommended. In view of the present level of the repeatabilities and reproducibilities (Table 1),an d experience in dairy cattle, it is concluded that accuracy of exterior scoring can achieve a satisfactory level (rj>0.6) for most traits if inspectors are intensively and regularly trained. Principal component analysis was used to determine whether exterior traits couldb e grouped together intoa fe wne wan d 'interprétable' traits (factors).Th eadvantag eo fprincipa lcomponen tanalysi s referst oth eles s comprehensive analysis of the data (e.g. in the case of breeding value

101 estimation)du e to thereduce dnumbe r ofne w traitswhic h are uncorrelated with each other,whil e using information onal l traits. Itwa s shown that 20exterio r traits could be summarized by five interprétable factors that explained 57%o fth etota lvariance .A reductio no fth enumbe ro fvariable s from 20 to 5, involved discarding 43Ï of the total variance. A total of thirteenfactor swer eneede dt outiliz ea tleas t90 %o fth etota lvariance . Forpi gbreedin gcompanie swhic har einvolve d incollectin g thedata ,cost s play an important role in the final decision as towhic h traits should be scored for exterior breeding value estimation. Calculation of the factor scores implies thatal l traits shouldb e scored. It istherefor e concluded that the merit of using exterior factor scores for selection purposes is questionable.

Heritabilitieso fexterio rtrait san drelationship swit hprowt hperformance .

An excellent growthperformanc e isonl yprofitabl e if time of delivery of the animal is achieved without (severe) constitutional problems manifested indevian tbehavio r ormalfunctio n ofleg s andbody . Because of thecontinuou sselectio nfo rgrowt hperformanc e inpi gbreeding ,correlate d responsesfo rexterio rtrait smigh tbecom erelevant .Thos eresponse s depend ongeneti c correlations and standarddeviations . Inchapte r 3,th e genetic background of exterior traits and their correlations with growth perfor­ mance were reported. Exterior traits describing body size and structure (trait 1 to 6) are moderately heritable except for length of the hams (h2-=0.06).Tha t trait isals o found tohav e arelativel y low repeatability and reproducibility (Table 1). It is therefore likely that the low heritability is due to inaccurate appraisal, probably because of short­ comings in the definition and description of the trait.Heritabilitie s of size (objectively measured)ar e generally reported in the range of 0.4 to 0.5 (e.g.Falconer ,1981) .Selectio nfo rsiz eshoul dtherefor eno tb ebase d onvisua l appraisal but on objectivemeasurements .

Of the forelegs, only side view of the pastern shows a significant heritability estimate.Heritabilit y estimates of the posture of rear legs range from 0.22 to 0.30, and are larger thanestimate s of individual rear legtrait sreporte db yWeb be tal . (1983).Differenc e insiz ebetwee n inner andoute rcla wi sconsidere d tob ea predisposin gcaus eo flamenes s (Penny,

102 1979).Th elo wheritabilit yfoun di nou row ninvestigation san db yWeb be t al. (1983)indicat e thatselectio no nsiz erati oo finne ran doute rcla w willno tgiv emuc hrespons ean di stherefor eirrelevan ti nnucleu sbreeding . Inpractice ,th ebreedin gvalu eo fgrowt hperformanc eo fnucleu spig s ofsir ean dda mline s isgenerall yestimate db y thetraits :dail ygrowt h (ADG), backfat thickness (UB),an d daily feed intake (DFI), using a selection index. Depending on correlations and heritabilities, visual appraisal of exterior traits can increase accuracy of the estimation. Relative to a performance index including ADG,U B and DFI as index and breeding goaltrait susin geconomi cvalue s0.178 ,-0.41 5an d-0.0 5 (Knap et al., 1985), respectively, thecorrelatio nbetwee n index andbreedin g goalincrease s7 %(fro m0.4 9 to0.52 )whe nsi xexterio rtrait sdescribin g sizean dstructur eo fth ebod y(trait s1 t o6 )ar eadde dt oth eindex .Th e exteriortraits ,regularit yo fbac kan dloin ,lengt han dwidt ho fth eham s account for 88% of this increase. The increase in accuracy of growth performance estimation by including information on exterior traits, however,decline swhe nperformanc einformatio no frelative si sinclude di n thebreedin gvalu eestimation .Schaeffe ran dKenned y(1986 )showe dtha tth e applicationo fa nanima lmode l forbreedin gvalu eestimatio n inpig swa s possibleo na Persona lComputer .I tis ,therefore ,likel ytha tpi gbreedin g companieswil linclud einformatio no fal lrelative so fa pi gfo rbreedin g valueestimation .I ti sconclude dtha tcontributio no fexterio rtrait sfo r growthperformanc eestimatio ni ssmal lan di sexpecte dt odecreas ei nth e near future,s oexterio r traitsshoul db econsidere d ofmino r importance forgrowt hperformanc eestimation . Inth ebreedin g goal,however ,n oattentio ni spai d totyp e (shapeo f thehams )o fth ecarcass ,wherea sclassificatio no fcarcasse saccordin gt o EECregulation s iso nth ebasi so flea nmea tpercentag ean d 'type'.Typ e therefore represents an economic value. No type measuring technique is developedye tfo rpractica lapplication ,s oclassificatio no ftyp ei sstil l by visual appraisal. Scoring the shape of the hams of live animals by visual appraisal might therefore be relevant in selection programmes. Furtherresearc h isneede d todetermin eth erelevanc eo fvisua lappraisa l oftyp eo flivin ganimal sfo rpredictio no fcarcas styp eclasses .

103 Exteriortrait san dlongevity .

Manyfactor saffec tth edecisio no fa herdsma nwhethe ro rno tt ocul l a sow and so affect the animal's longevity. Some of these factors are reproductiveperformance ,constitution ,diseases ,age ,character ,an dth e priceo fgilt san dsows .I ti sobviou stha tth edisposa ldecisio ni smainl y based on characteristics of the sow that are directly related to the animal'sfunctionin gan dperformance .A sstate db yBrod y (1945),functio n isintimatel y correlatedwit hstructur ean dform .Exterio r traits,whic h describestructur ean dfor mo fa nanima lar etherefor eo finterest . The mean annual culling rate for sows from different studies varies between 39an d 55%,wit h themai nreason s fordisposa lbeing :fertilit y problems (40%),lo wnumbe ro fpiglet s(30% )an dle gweaknes s(15% )(Dagor n andAumaitre ,1979 ;Pattison ,1980 ;T eBrake ,1986 ;Va nSteenberge ne tal , 1990). Disposal because of leg and foot disorders causes the largest economic loss (Dijkhuizene tal. , 1989).Whe nle gan dfoo tdisorder sca n betimel yforeseen ,the nthi slos sca nb ereduce dthroug hmor eflexibilit y inth echoic eo f timet ocul lth eanimal .Reductio no fdisposal sbecaus e ofle gweaknes swil lincreas ereproductiv elifetim ean dmigh tb eachieve d byearl yselectio no nexterio rtrait swhic har erelate dt ole gweakness . Theeconomi crelevanc eo freproductiv elifetim eo fsow si sdescribe db y Kroesan dVa nMal e (1979),Dijkhuize ne tal .(1986 )an dD eVrie s (1989). Anincreas eo faverag elongevit yfro mfou rt ofiv elitter sha sa neconomi c benefit of approximately Dfl. 45.-pe r sowpe ryea r (Dijkhuizene t al., 1986;D eVries ,1989) . Constitutiono fsow si so finteres tfo rnucleu sda mline san dcommercia l multiplierherds .I nchapte r4 ,eigh texterio rtrait sjudge da tgil tstag e werefoun dt ob esignificantl yrelate dt odisposa lbecaus eo fle gweakness . Those traits are:lengt han dwidt ho fbac kan d loin,widt h of thehams , sidevie wo fth erea rlegs ,knottines so fth erea rlegs ,gai t(movements) , swingingo fth ebac kan dth elifetim eexpectatio nscore .Selectio ni nda m lines is for a combination of growth and reproductive (fertility and longevityo fsows )performance .I nvie wo fth egeneti ccorrelation sbetwee n exteriortrait san dgrowt hperformance ,an drelationship sbetwee nexterio r traits and disposal because of leg weakness, it can be concluded that selection for growth performance negatively influences leg weakness by

104 increasingwidt ho fback ,loi nan dhams ,slowin gdow nth egai t(locomotor y ability) and increased swinging of the back. These exterior traits thereforedeserv eattentio ni npi gbreeding . Calabottae tal . (1982)state dtha tlongevit yi na breedin gherd ,whic h amongothe rfactor si sth eresul to fal lknow nan dunknow nmanifestation s ofle gweakness ,migh tb ea goo dindicato ro fle gweakness .I nchapte r5 relationshipsbetwee nexterio r traitsjudge da tgil tstag e andlongevit y wereanalyzed .Significan tlinea rregressio ncoefficient so ffou rexterio r traits on thenatura l logarithm of longevity are found: shoulderblade , gait(movements) ,swingin go fth ebac kan dth elifetim eexpectatio nscore . Significant quadratic regression coefficients of longevity on exterior traitsar efoun d for:widt ho fbac kan dloin ,fron tan dsid evie wo fth e forelegs,sid evie wpaster no fth eforelegs ,sid evie wrea rleg san dsid e viewo fth ehoc kjoint .Extreme so fthes etrait sar eunfavourabl yrelate d tolongevity .Th eeffec to fmodificatio no fexterio r traitso nlongevit y isfoun dt ob erelevan tfo rth econnectio no fth eshoulde rblad ewit hth e body, width of the back, loin, and hams, sidevie w of the hockjoint , quantityo fliqui d(swelling )a tth ehoc kjoint ,gai t(movements) ,swingin g ofth ebac kan dlifetim eexpectation .Fo rmultiplie rfarm sthes etrait sca n beuse dt oestimat e thegilt' sfutur eprofitability .Base do naccurac yo f themeasurement san dth eeffec to fth etrait so nlongevit y (Table 1), the fivemos t interesting exterior traits formultiplie r farms, inorde ro f decreasing relevance are: (1)widt h ofhams , (2)gai t (movements), (3) lifetime expectation, (4)connectio no fth eshoulde rblad ewit h thebod y and (5)swellin ga tth ehoc kjoint . The effect of selection for longevity in dam lines of pig breeding companies depends on genetic correlations between exterior traits and longevity.Thes ecorrelation scoul dno tb eestimate di nth epresen tstud y andwer eals ono t found inth e literature.Value so fphenotypi ccorrela ­ tions might however, be useful estimates for the genetic correlations. Inclusioni na selectio ninde ximplie sa know neconomi cvalu ean dlinearit y ofthi svalue .Severa lle gtrait sshowe da noptimu mfo rlongevit yan dar e therefore troublesome to include into a selection index. Because the economicvalu eo fexterio r traits isa sye tno tquantifie d andlinearit y ofth eeconomi cvalu eo fsevera l traits isdoubtful ,independen tcullin g

105 on exterior traits that are related to longevity, might be an attractive alternative for index selection.

Main conclusions.

From the investigations presented in this thesis, the following main conclusions canb e draw: Asatisfactor yaccurac yo flinea rscorin go fexterio r traitsi npig sca n be achieved. When judgements are made by several persons, regular deliberation and training are recommendable. Exterior traits describing size and structure of the body are not relevant for growthperformanc e estimation. Selection only on growth performance has an unfavourable effect on locomotory ability ofboars . Broad back, loin and hams at gilt stage are unfavourably related with sow lifetime.A n increased herd life,whic h iso f interest for multiplier farms, might be achieved by an increased exterior selection in gilts. A broad body, slow gait (movements) and extreme leg postures are the main exterior criteria onwhic h gilts shouldb e selected out. Selection for constitution in dam lines should be concentrated on the exterior traits: gait (movements) and body width, whereas animals with extremepostur eo fth elegs ,especiall ybuckle dknees ,bowe dan dstee p rear legs, shouldb e discarded forbreedin gpurposes .Fo r a long lifetime,pig s with an easy and quick gait, and a relatively narrow sized body are desirable.

REFERENCES

Brody, S., 1945. Bioenergetics and growth. Reinhold Publ. Co., New York, 1023pp . Calabotta,D.F. , Kornegay, E.T.,Thomas ,H.R. ,Knight ,J.W. , Notter,D.R ., Veit, H.P., and Wright, M.E., 1982. Restricted energy intake and elevatedcalciu man dphosphoru s intakefo rgilt sdurin ggrowth .II .Gai t characteristics analyzed from 16mmmotio npictur ephotography .J . Anim. Sei. 55:1395-1404 .

106 Dagorn, J. and Aumaitre,A. , 1979. Sow culling: Reasons for and effect on productivity. Livest. Prod. Sei. 6: 167-177. De Graaf, Hamoen, A. and Koorn, D.S., 1986. Type classification in the Netherlands.Roya lDutc hCattl eSyndicat e (K.N.R.S.),86-678/FG/AP ,1 0pp . DeVries ,A.G. , 1989.A model toestimat e economicvalue s oftrait s inpi g breeding. Livest. Prod. Sei. 21:49-66 . Diers, H. and Swalve, H., 1987. Estimation of genetic parameters and breeding values for linear scored type traits. 38th Annual Meeting of the EAAP, Portugal,abstrac tG5a20 . Dijkhuizen,A.A. ,Morris ,R.S .an dMorrow ,M. , 1986.Economi c optimization of culling strategies in swine breeding herds, using the "PorkCHOP computer program". Prev.Vet .Medic , 4:341-353 . Dijkhuizen, A.A., Krabbenborg, R.M.M, and Huirne, R.B.M., 1989. Sow replacement: a comparison of farmers' actual decisions and model recommendations. Livest. Prod. Sei. 23:207-218 . Falconer, D.S., 1981. Introduction to quantitative genetics. Oliver and Boyd, Edinburgh, London, 340pp . Knap,P.W. ,Huiskes ,J.H . andKanis ,E. ,1985 .Selectio n index for central test in Dutch pig herdbook breeding from 1984.Livest . Prod. Sei. 12: 85-90. Kroes, Y. and Van Male, J.P., 1979. Reproductive lifetime of sows in relation to economy ofproduction . Livest. Prod. Sei. 6: 179-183. Murphy, D.T. and Sinnott, M.L., 1989.Relationship s between progeny test results for linear type,bee f andmil kproductio n inFriesia ncattl e in Ireland. 39th Annual Meeting of the EAAP,Dublin ,pape r C3.12, 9pp. Pattison,H.D. ,1980 .Pattern so fso wculling .Pi gNew san dIn fo .1 :215-218 . Penny, R.H.C., 1979. Genetical, physiological and anatomical factors contributing to foot and limb disorders in growing and adult pigs, including a statistical review of foot and limb disorders in pigs attributable to floors. Proceedings PigVet . Soc. 4:85-96 . Schaeffer,L.R . andKennedy ,B.W. ,1986 .Computin gsolution s tomixe dmode l equations. 3rdWorl dCongres sGen .Appl .t oLivest .Prod. ,Nebrask aXII : 382-393. Te Brake, J.H.A., 1986. Culling of sows and the profitability of piglet production. Neth. J. Agric. Sei. 34:427-435 .

107 Thompson,J.R. , Freeman, A.E., Wilson, D.J., Chapman, C.A. and Berger, P.J., 1981.Evaluatio no fa linea rtyp eprogra m inHolsteins .J .Dair y Sei.64 :1610-1617 . VanSteenbergen ,E.J. ,Kanis ,E. ,Koops ,W.J .an dVa nde rSteen ,H.A.M. , 1990.Exterio ri nsows :1 .Effec to fparit ynumbe ran dassociatio nwit h reasonsfo rdisposal .Livest .Prod .Sei .(submitted) . Webb,A.J. ,Rüssel ,W.S .an dSales ,D.I. ,1983 .Genetic so fle gweaknes s inperformance-teste dboars .Anim .Prod .36 :117-130 .

108 SUMMARY

Infar manimal scharacterizatio no fexterio r isno ta goa l initsel fbu t rather serves as an indicator of economically important traits, i.e. for pigs:growt hperformance ,reproductivit y andconstitution .Thi s indication might be of interest when these traits can not be measured on living animals (e.g. slaughter and meat quality) or when traits can only be measured on one sex or at later age (e.g.pigle t production, longevity). Severalmethod sar eavailabl et ocharacteriz eexterior :direc tmeasurement s (e.g.width ,height ,circumference) ,photograph yan dvisua lappraisal .Dis ­ advantages of the last method are subjectivity and the relatively low accuracy of the characterization. Despite of those disadvantages visual appraisal is frequentlyuse dmainl ybecaus eo fth elo wcost san d quickness ofmeasuring . Since introduction of ultrasonic measuring equipment, prediction of slaughter characteristics of living pigs can be done more accurately by ultrasonic backfat measurement then by visual appraisal. Since then, the main relevance for appraisal of exterior traits is due to presumed relationships with constitution and reproduction. Leg and foot disorders are a main constitutional problem in pigs. These disorders are generally describedb y the term legweakness .Th eproportio n of testednucleu s boars culled for leg weakness varies from 15 to 50Z. In sowherd s 15 to 20%o f the disposed sows are culledbecaus e ofle gweakness .Economi c loss due to legweaknes s can arise indifferen t ways: (1)reduce d genetic improvement in performance traits at nucleus level due to involuntary culling, (2) premature disposal ofbreedin g stocka tnucleu san dmultiplie r levels, (3) loss of sales at nucleus andmultiplie r levels,an d (4)reduce d reproduc­ tion and production efficiency at all levels. Next to the economic consequences, leg weakness has a negative impact on animal welfare and emphasizes the negative image of intensive husbandry systems.

Inpi g breeding programmes emphasis is given toexterio r traits mainly as ametho d to select for abette r constitution. Very little,however , is known about the effectiveness of that selection. Responses to selection depend on heritabilities of exterior traits and on correlations between exterior traits andconstitution ,wherea s applicationo fmoder n techniques

109 forbreedin gvalue sestimatio nnecessitate sa nappropriat ewa yfo rexterio r scoring.

The aim of this study was to evaluate the importance of exterior appraisal in pig breeding. In the first chapter the function of exterior judgement inpi gbreedin g and husbandry isreviewed . Special attention is given to relationships between exterior and the problem of foot and leg disorders.

Investigations were set up to obtain the parameters that are relevant forth eevaluatio no fth eimportanc eo fexterio rappraisa l inpi gbreeding . Before investigations started, exterior traits to be investigated were chosen indeliberatio nwit h inspectors of theDutc hpi gherdboo k and three pigbreedin g companies.Th e choice of the traits tob e evaluatedwa sbase d on the expected economic value and the likelihood to have a relationship with constitution. Results and conclusions are based on appraisal of exterior traits of centrally tested boars and of gilts and sows on 24 commercial multiplier farms.

Methods of exterior appraisal.

A large number of exterior appraisals is needed for a sufficient reliability of heritability and correlation estimates. Ease, and time necessary for a judgement are therefore important criteria for the choice of themetho d for exterior characterization. Visual appraisal meets those criteria.I nliteratur esevera ladvantage sar ementione d forlinea r scoring of visually appraised exterior traits. For present investigations 19 distinct and two composite exterior traits are defined and described on a linearscal ewhic hrange d from0 t o9 wit h0. 5 intervals.Trait s considered refer to size and structure of thebod y (6),foreleg s (3),rea r legs and claws (8),movement s (2),an d alifetim e expectation scorewhic h should be anestimat e ofth eanimal' s life timea sinterva lbetwee ntim eo fjudgemen t and expected disposal. The evaluation of the linear scoring is described inchapte r 2.

The linear scoring system isevaluate d interm sof :difference s between inspectors,us eo fth e1 9 (0t o9 wit h0. 5 increments)categories ,accurac y

110 ofscorin gan drelationship sbetwee ntraits .Fo rtha tpurpos e4 0boar swer e judged twice by each of 10 inspectors. Differences in mean and standard deviationfo rlinearl y scoredexterio r traitsoccurre dbetwee n inspectors, with most inspectors showing a marked preference for whole number categories. On average, repeatability was 0.60 and reproducibility, indicating thecorrelatio nbetwee nrepeate dmeasurement sdon eb y different inspectors, was 0.30. Exterior traits could be summarized by principal component analysis to five new independent exterior variables, that are characterized as 1) side view of legs, 2) way of walking and lifetime expectation, 3)size ,4 )harmon y of thefram e and 5)rea rleg s andclaws . Inpresen t form, usefulness of application of described linear scoring system for exterior traits for selection purposes highly depends on the persons who judge the animals. When selecting and training appropriate people, a satisfactory scoring accuracy canb e achieved for mosttraits .

Heritabilitieso fexterio rtrait san drelationship swit hgrowin gperformance .

In chapter 3 genetic relationships between growth performance and 20 linearly scored exterior traits are estimated from data of 2792 boars of severalstrains .Boar sar eteste dwit ha dlibitu mfeedin gfro mapproximatel y 25 to 105 kg at central test stations of the Dutch herdbook and three breeding companies.Heritabilit y estimates of exterior traits ranged from 0.01 to 0.38. Traits describing body size and harmony (trait 1 to 6) are moderately heritable (h2-0.15-0.3)excep t length of theham s (h2=0.06). Of theforelegs ,onl ysid evie wo fth epaster nshow sa significan theritabilit y estimate. Heritability estimates ofpostur e of rear legs ranged from 0.22 to 0.30. Movements and constitution score have a small genetic component

2 (h -0.15). Close genetic correlations (rg=0.4-0.5)ar e foundbetwee n daily gain and side view of foreleg pastern, side view and dryness of the hock joint. Slower growing animalshav e steeper forelegpastern ,les s liquid at the hock joint and more sickle hocked rear legs. The genetic correlation between feedconversio nan drea rvie wo fth erea rleg sindicate s thatboar s withbette r feed conversion have more 0-shaped rear legs.

Atwo-stag e selectionstrateg ywit hselectio ni nth efirs tstag eo ngai t pattern (movements) only and on growth performance in the second stage, reduces genetic improvement of performance traits.Whe n assuming a nihil

111 economicvalu efo rgai tpatter nan drejectin g2 5o r50 %o fth eboar si nth e firststage ,the neconomi cmeri ti sreduce db y1 3o r24% ,respectively . The accuracy of the estimatedbreedin g value for growthperformance , based on an own performance indexwit h the index traits:dail y growth, backfat thickness, and daily feed intake, shows a minor increase when exteriortrait sdescribin gsiz ean dstructur eo fth ebod yar einclude di n theindex .

Exteriortrait san dconstitution .

Inpractic emos tinteres tfo rexterio rtrait si sbecaus eo fth eexpecte d correlations between exterior traits and longevity as determined by the animals constitution. Over a two-year period, several thousand gilts between7 an d9 month so fag ean dsow sfro mweanin gtil labou t3 0day spos t weaning,wer erepeatedl yjudge d forexterio ro n2 4commercia lmultiplie r herds.I nchapter s4 an d5 relationship so flinearl yscore dexterio rtrait s withreason sfo rdisposa lan dlongevit yar estudied .Som eexterio rtrait s judgeda tgil tstage ,showe dt ob erelevan tfo rdisposa lfo rle gweakness . Sowsculle dfo rle gweaknes sbefor e4t hparit ywer elonge ran dbroade ra t gilt stage,ha d less straight rear legs and more tubercles at the rear legs,the ywalke dslowe ran dtwiste dmor ewit hth ehin dquarter scompare d toanimal stha tha dproduce da tleas t4 litters .Sow sculle dbefor efourt h paritybecaus eo fle gweakness ,ha da lowe rscor efo rconstitutio na tgil t stagetha nsow stha tproduce d> 4 litters . Formos ttrait sconcernin g thepostur eo fth elegs ,bot h lowan dhig h scoreswer enegativel y correlatedwit h longevity.Th eeffec to fa chang e ofon estandar d deviationuni t (s.d.)fro m theaverag eexterio r scoreo n length of productive life (LPL), as a measure for longevity, is con­ siderable for a number of exterior traits. Broader hindquarters are unfavourablyrelate dt oLP L(-5 3 dayspe rs.d. )whil ewit hon es.d .thinne r hindquartersLP Lincrease s4 4days .LP Lincrease swit hdecreasin gswellin g at the hock joint. A more flexible gait pattern, less twisting of the hindquartersan da highe rlifetim eexpectatio nscor ecoincid ewit ha highe r LPL (67,4 4 and 42 days per s.d. respectively). Selection on exterior traitswit hmai nattentio nfo rgai t(movements )an dbod ywidt hca nincreas e

112 average length of productive life and so decrease the annual involuntary disposal of sows.

Mainconclusions .

From the investigations presented in this thesis, the following main conclusions canb e draw: Asatisfactor y accuracyo flinea rscorin go fexterio rtrait si npig sca n be achieved. When judgements are made by several persons, regular deliberation and training are recommendable. Exterior traits describing size and structure of the body are not relevant for growth performance estimation. Selection only on growth performance has an unfavourable effect on locomotory ability ofboars . Broad back, loin and hams at gilt stage are unfavourably related with sow lifetime.A n increasedher d life,whic h iso f interest for multiplier farms, might be achieved by an increased exterior selection in gilts. A broad body, slow gait (movements) and extreme leg postures are the main exterior criteria onwhic h gilts shouldb e culled. Selection for constitution in dam lines should be concentrated on the exterior traits: gait (movements) and body width, whereas animals with extremepostur eo fth e legs,especiall ybuckle dknees ,bowe dan dstee prea r legs,shoul db e discarded forbreedin gpurposes .Fo ra long lifetime,pig s with an easy and quick gait, and a relatively narrow sized body are desirable.

113 SAMENVATTING

Exterieur beoordelen bij landbouwhuisdieren is geen doel op zichzelf, maarword t gebruikt tervoorspellin gva neconomisc hbelangrijk ekenmerken . Bijvarken szij nproduktiekenmerke nzoal sgroei e nspekdikte ,reproduktie - kenmerken en constitutie economisch gezien belangrijk. Onder constitutie wordt de algehele lichamelijke gesteldheid van het dier verstaan. De voorspellende waarde van het exterieur wordt met name dan van belang wanneer dekenmerke nwaa r het in feiteomgaa tnie t direct aanhe t levende dier gemetenkunne nworde n (b.v. slacht- envleeskwaliteit ) ofwannee r de kenmerkenpa s op latere leeftijd gemetenkunne nworde n (b.v.levensduu r of -produktie). Voor het karakteriseren van het exterieur zijn diverse methoden beschikbaar:direc tmete n (b.v.lengte) ,fotografere no ffilmen ,e nvisuee l beoordelen. Op het oog beoordelen is een subjectieve methode waarvan de nauwkeurigheid vaak matig is. Desondanksword the tvisuee l karakteriseren van exterieurkenmerken veelvuldig toegepast. De voordelen van de methode zijnd elag ekoste ne nd esnelhei dwaarme ed ewaarneminge nverkrege nworde n alsmede depraktisch e toepasbaarheid. Sinds de ontdekking dat ultrasone geluidstechnieken gebruikt kunnen worden om de spekdikte bij levende varkens nauwkeurig te meten, kan de slachtkwaliteit nauwkeuriger aan het levende dier worden gemeten door ultrasone spekdikte-metingen dan door visuele beoordelingen van het exterieur.Sindsdie nconcentreer texterieurbeoordelin g zichme tnam eo p de veronderstelde relaties met constitutie en reproduktie. Een belangrijk constitutie-probleem bij varkens heeft betrekking op beengebreken. Zowel in de varkensfokkerij als de -houderijword t een aanzienlijk deelva n de dierenvoortijdi gafgevoer dvanweg eonvoldoend efunctioneren dbeenwerk .He t economisch verlies wordt veroorzaakt door: (1) verminderde genetische vooruitgang inproduktiekenmerke n op topfokniveau, (2)voortijdig e afvoer van zeugen opvermeerderings - en topfokbedrijven, (3)minde r fokmateriaal beschikbaar voor verkoop aan zeugenbedrijven, en (4)verminderd e (repro­ duktie-efficiënti eva nzeugen ,bere ne nslachtvarkens .Naas td eeconomisch e gevolgen van beengebreken zijn ook de welzijnsaspecten van belang. Veel beenwerkproblemenzij nongezon dvoo rzowe ld evarken sal sd evarkenssector .

115 In de huidige varkensfokkerijprogramma's wordt het exterieur voor­ namelijkgebruik tal svoorspelle rva nconstitutie .D egenetisch evooruit ­ gangvoo rconstituti ehang ta fva nd eerfelijkhei dva nexterieurkenmerke n enva nd ecorrelatie s tussenexterieu re nconstitutie .Voo rhe tschatte n vanfokwaarde svoo rexterieu rkenmerke ni see ngeschik tbeoordelingssystee m nodig.

Hetdoe lva ndi tonderzoe ki sd ebepalin gva nhe tbelan gva nexterieur ­ beoordeling in de varkensfokkerij. In het eerste hoofdstuk wordt een literatuuroverzichtgegeve nva nd erelatie sva nexterieu rme t(re)produk - tiekenmerkene nconstitutie .Daarbi jword tvoora laandach tgeschonke naa n derelati etusse nexterieu re nbeenwerkproblematiek .

Een aantalpraktijkonderzoeke n zijn opgezet om devoo r dit onderzoek relevantekengetalle nbove nwate r tekrijgen .O p2 4zeugenbedrijve nzij n gedurende tweejaa r geltene nzeuge no pexterieu rbeoordeeld ,terwij lo p selectiemesterijen en toetsstallen van varkensfokkerij-instellingen gedurende een jaar enkele duizenden getoetste beren op exterieur zijn beoordeeld.Voorafgaan d aanhe texterieurbeoordelen ,zij ni noverle gme t debetrokke nbeoordelaar sd et ebeoordele nkenmerke nvastgesteld .D ekeuz e vand ekenmerke nwer dvoornamelij kbepaal ddoo rd everwacht e relatieme t constitutie.

Beoordelingsmethodieken

Voorhe tverkrijge nva nvoldoend enauwkeurig ekengetalle nzij nee ngroo t aantal exterieurbeoordelingen noodzakelijk. Gemak en snelheid van beoordelenzij nda nbelangrijk ecriteri avoo rd ekeuz eva nd ebeoordelings ­ methode. Visuele beoordeling voldoet aan die criteria. Uit statistisch oogpunte nvoo rd einterpretati eva nd eresultate nbied the tbeoordele nva n enkelvoudige kenmerken welke gescoord worden op een lineaire schaal, duidelijkevoordele nbove nander evisuel ebeoordelingssystemen .Daaro mzij n voor dit onderzoek exterieurkenmerken zoveel mogelijk enkelvoudig gedefinieerde nbeschreve no pee nlineair eschaa lwelk eloop tva n0 to t9 met h intervallen.Ze skenmerke nhebbe nbetrekkin g opd elichaamsbou we n -omvang, drie kenmerken beschrijven de stand van de voorpoten, terwijl

116 achterpotene nklauwe nworde ngekarakteriseer d doorach tkenmerken . Verder wordenbeoordeeld :mat eva nzijwaart sbewege nva nd eachterhan dtijden she t lopen, de soepelheid van het lopen (gangen), kwaliteit en grootte van de spenene nverwacht e levensduur, tenotere nal she taanta lverwacht eno g te producerenworpen . Debruikbaarhei d vanhe t lineaire beoordelingssysteem isbeoordeel d op grond van: persoonsverschillen, gebruik van de 19 schaalklassen, nauw­ keurigheid van beoordelen (herhaalbaarheid) en de correlaties tussen de beoordeeldekenmerken .Daarto ezij n4 0bere nva nongevee r 100k go péé nda g twee keerbeoordeel d door 10beoordelaars . Erwerde n duidelijke persoons­ verschillen in gemiddelde en spreiding gevonden. Ook werd bij de meeste personen een duidelijke voorkeur voor het gebruik van gehele getallen geconstateerd. De gemiddelde herhaalbaarheid binnen personen is 0.60 en tussenpersone n 0.30. Hetme tbehul p van factor analyse reducerenva n het aantal kenmerken door kenmerken samen te voegen tot een nieuw kenmerk, resulteerde in de volgende onderling onafhankelijke nieuwe kenmerken: 1) zijaanzicht van de poten, 2)bewegingen/levensduur , 3)groott e en omvang, 4)lichaamsbouw , en 5)achterpote n enklauwen . De praktische bruikbaarheid van het lineaire beoordelingssysteem voor exterieur is duidelijk afhankelijk van de beoordelaar. Wanneer de juiste mensen worden gekozen en wanneer zij regelmatig worden getraind en begeleid,da nka nd ebruikbaarhei dva nhe tsystee mal sgoe d geclassificeerd worden.

Erfelijkheid van exterieur kenmerken en relaties met produktie

In hoofdstuk 3 zijn de erfelijke achtergrond van exterieur en de relaties met produktiekenmerken geanalyseerd in materiaal dat afkomstig wasva ntoetsstalle nva ndivers e fokkerij-instellingen.D eresultate n zijn gebaseerd op gegevens van 2792 onbeperkt gevoerde en voor het merendeel individueel gehuisveste beren uit diverse lijnen. Llchaamsgrootte is middelmatig erfelijk (h2-0.15-0.30).D eerfelijkheidsgraa dva n individuele achterpoot-kenmerken varieert van 0.22 tot 0.30. Gangen en algehele gesteldheidzij nmati gerfelij k (h2~0.15).Duidelij kgenetisch e correlaties

tussengroe ie nenkel eexterieurkenmerke n(r g=0.4-0.5)geve naa nda ttrage r groeiende dieren aanleg hebben voor steiler voorpootkoten, droger hakken

117 ensabelbenig e achterpoten.Diere nme tee nhog evoederefficiënti ehebbe n aanlegvoo rO-benlg eachterpoten . Het meenemen van exterieurbeoordelingen voor lichaamsmaten in de fokwaardeschattingvoo rproduktieaanle ggeef tslecht see ngering everhogin g vand enauwkeurighei dva ndi eschatting .He tbeoordele nva ndi eexterieur - kenmerkenvoo rhe thiervoo rgenoemd edoe llijk tdaaro mnie tzinvol . Beenwerkproblemen zullen zich het eerst manifesteren inhe t kenmerk gangen.Ee nselecti eo phe tkenmer kgange nme tal sdoe lhe tvermindere nva n de beenwerkproblemen blijkt gevolgen te hebben voor de genetische vooruitgang in de produktiekenmerken. Een voorselectie van toetsberen waarbij 50% van de dieren afgekeurd wordt op gangen vermindert de genetischevooruitgan gvoo rd eproduktiekenmerke nme t24X .

Exterieurkenmerkene nconstituti e

In de praktijk is de belangstelling voor het varkensexterieur voor­ namelijkhe tgevol gva nd everondersteld erelatie stusse nexterieu re nd e doord econstituti ebepaald elevensduur .He tonderzoe knaa rdez erelatie s isgebaseer d opexterieurbeoordelinge n aanenkel eduizende ngelte no p2 4 zeugenbedrijven.He tovergrot edee lva nd ebeoordeeld egelte ni saangekoch t opee nleeftij dva nze smaanden .Ee naanzienlijk evoorselecti eo pexterieu r heeft danwaarschijnlij k reedsplaat s gevonden.I nhe tvierd e envijfd e hoofdstuk zijnd egevonde nrelatie sva nexterieu rme tredene nva nuitva l enlevensduu rbeschreven .Enkel eexterieurkenmerke nbeoordeel daa ngelte n vertonen een verband met uitval op grond van beengebreken. Zeugen uitgevallenvóó rd evierd ewor pvanweg ebeengebreken ,ware nal sgei tte n opzichteva nd e geltendi e laterbeweze nhebbe nminimaa lvie rworpe nt e kunnenproduceren ,lange re nbreder ,hadde nminde rsteil eachterpoten ,e n haddenee ngrote rdee lva nd eachterpote nbedek tme tknobbels .Hu ngange n waren trager terwijl de achterhand meer heen en weer bewoog, en de beoordelingva nd ealgehel egesteldhei dwa sminder . Bij de meeste beenwerkkenmerken waren zowel hoge als lage scores (extremen) ongunstig gerelateerd met levensduur. Afwijking van het gemiddelde exterieur met 1 eenheid spreiding (s.d.), heeft bij diverse exterieurkenmerken een duidelijk effect op de levensduur. Een matig aangeslotenschoude ri snegatie fgerelateer daa nlevensduur .Bred ehamme n

118 zijnnegatie fgecorreleer dme tlevensduu r (-53 dagenpe r s.d.). terwijlbi j 1 s.d. smallere hammen de levensduur gemiddeld 44 dagen langer is. De levensduur neemt toeme the t droger wordenva n dehakken . Soepele gangen, weinig zijwaartsbewegend e achterhand enee nhoger e scorevoo r de algehele gesteldheid zijn gunstigvoo r eenhog e levensduur (respectievelijk 67,4 4 en4 2 dagenpe r s.d.). Ondanks devermoedelijk e voorselectie op exterieur op zes maanden leeftijd, blijkt de bij diverse kenmerken nog aanwezige variatie voldoende te zijn om duidelijke verschillen in levensduur aan te kunnen tonen.

Belangrijkste conclusies

Uit ditproefschrif t kunnen devolgend e conclusies getrokkenworden : - Lineaire beoordeling van het exterieur is praktisch goed uitvoerbaar. Wanneerd ejuist emense nworde ningeschakel dvoo rd eexterieurbeoordelinge n en wanneer zij regelmatig worden getraind en begeleid, dan is de nauw­ keurigheid van debeoordelinge n acceptabel.D e correlatiestructuur tussen deexterieurkenmerke n iszodani g dathe t samenvoegenva nkenmerke n tot een kleiner aantal nieuwe kenmerken niet isaa n tebevelen . -Beoordelin gva nd e lichaamsbouw en -omvang tenbehoev eva n de fokwaarde­ schatting voor produktiekenmerken isnie t relevant. -Ee nbred e midden- en achterhand als geit zijn ongunstig gerelateerd aan levensduur. Een verhoging van de gemiddelde levensduur van zeugen kan bereikt worden door selectie op exterieur bij gelten. Brede dieren met trage gangenkome n als eerste inaanmerkin g om afgevoerd teworden . -Selecti e op levensduur inzeugenlijne n dient zich te concentreren op de kenmerken: gangen en lichaamsbreedte, terwijl dieren met extreme poot standen, in het bijzonder bokbenige voorpoten, onderstandige en steile achterpoten, niet voor de fokkerij moeten worden ingezet. Voor een hoge levensduur zijnnie t tebred e dierenme t soepele gangen gewenst.

119 Curriculumvlta e

Evert Jan van Steenbergen werd op 26 januari 1959 geboren teBenneko m (gemeente Ede),waa r hij op een veehouderijbedrijf opgroeide. In 1977 behaaldehi jhe tVW Odiplom aaa nhe tChr .Streeklyceu mt eEde .I ndatzelfd e jaarbego nhi jme tzij nstudi eZoötechnie kaa nd etoenmalig eLandbouwhoge ­ school teWageningen .He tdoctoraa lexame nwer dafgeleg d injanuar i198 4 met als hoofdvakken de Veeteelt en de Gezondheids- en Ziekteleer der Huisdierene nal sbijva kAgrarisch eBedrijfseconomie .Va naugustu s198 3to t januari198 4wa shi jwerkzaa mal sdocen taa nd eChr .Middelbar eAgrarisch e School teBarneveld .Vana f 9januar i 1984wa shi j aangesteld alsweten ­ schappelijkmedewerke r intijdelijk e dienstbi jd evakgroe pVeefokkerij . Sinds5 maar t 1990i shi jwerkzaa mal sgeneticu sbi jd evarkensfokkerij - groeperingCofo kt eOosterhou t (Gld).

120