analysis of longevity in slovenian holstein cattle. acta agriculturae slovenica

8
Acta argiculturae Slovenica, 98/2, 93–100, Ljubljana 2011 doi:10.2478/v10014-011-0025-5 COBISS: 1.01 Agris category code: L01 ANALYSIS OF LONGEVITY IN SLOVENIAN HOLSTEIN CATTLE Klemen POTOČNIK 1 , Vesna GANTNER 2 , Jurij KRSNIK 1 , Miran ŠTEPEC 1 , Betka LOGAR 3 , Gregor GORJANC 1 Received August 20, 2011; accepted September 22, 2011. Delo je prispelo 20. avgusta 2011, sprejeto 22. septembra 2011. 1 Univ. of Ljubljana, Biotechnical Fac. Dept. of Animal Science, Groblje 3, SI-1230 Domžale, Slovenia 2 J.J. Strossmayer Univ. in Osijek, Fac. of Agriculture, Trg Svetog Trojstva 3, 31000 Osijek, Croatia 3 Agricultural institute of Slovenia, Hacquetova 17, SI-1000 Ljubljana, Slovenia Analysis of longevity in Slovenian holstein cattle e longevity of Slovenian Holstein population was ana- lysed using survival analysis with a Weibull proportional haz- ard model. Data spanned the period between January 1991 and January 2010 for 116,200 cows from 3,891 herds. Longevity was described as the length of productive life – from first calving till culling or censoring. Records above the sixth lactation were censored to partially avoid preferential treatment. Statistical model included the effect of age at first calving, stage of lacta- tion within parity, yearly herd size deviation, season defined as year, herd, and sire-maternal grandsire (mgs). Some effects had time varying covariates, which lead to 1,839,307 or on average 16 elementary records per cow. Herd and sire-maternal grand- sire effects were modelled hierarchically. Pedigree for sires and maternal grandsires included 2,284 entries. Estimated variance between herds was 0.12, while between sire variance was 0.04. Heritability was evaluated at 0.14. Genetic trend for sires was unfavourable, but not significant. A further research is needed to define the required number of daughters per sire and the dy- namics of genetic evaluation for sires whose majority of daugh- ters still have censored records. Key words: cattle / breeds / Slovenian Holstein / longevity / Weibull proportional hazards model Analiza dolgoživosti pri črno-beli pasmi goveda v Sloveniji Za analizo dolgoživosti smo pri slovenski črno-beli po- pulaciji govedi uporabili metodologijo analize preživetja in Weibullov model sorazmernih ogroženosti. V analizo smo vključili podatke 116.200 krav iz 3.891 čred skozi obdobje od januarja 1991 do januarja 2010. Dolgoživost je bila predstavlje- na kot doba produktivnega življenja, ki je definirana kot število od prve telitve do izločitve ali do datuma zajema podatkov za živali, ki so na ta datum bile še žive. Šesto in kasnejše laktacije smo okrnili na konec šeste laktacije, da smo omilili precenje- nost boljših živali. V statistični model smo vključili vpliv sta- rosti ob prvi telitvi, stadija laktacije ločeno za vsako zaporedno laktacijo, spreminjanje velikosti črede med leti, leto telitve, čre- do, očeta in materinega očeta. Ravni nekaterih vplivov so ča- sovno spremenljivi, kar povzroči, da smo v analizi obravnavali 1.839.307 zapisov ali povprečno 16 osnovnih zapisov na kra- vo. Čreda in vpliv očeta z materinim očetom sta bila v model vključena hierarhično. Rodovnik za očete in materine očete je obsegal 2.284 zapisov. Ocenjena varianca za vpliv črede je zna- šala 0,12, medtem ko je ocena variance med očeti znašala 0,04. Dednostni delež je bil ocenjen na 0,14. Genetski trend ima ne- gativno smer, a ni statistično značilen. Potrebne bodo nadaljnje raziskave, da bomo določili zadostno število hčera po biku in dinamiko obračunov plemenskih vrednosti za bike, ki imajo večino hčera še v fazi prireje. Ključne besede: govedo / pasme / slovenska črno-bela pa- sma / dolgoživost / Weibullov model sorazmernih ogroženosti 1 INTRODUCTION Longevity is a trait with great impact on dairy pro- duction economy and is, therefore, of considerable im- portance in dairy cattle breeding programmes (Charffed- dine et al., 1996; Strandberg and Soelkner, 1996). With the increase of longevity, the proportion of mature cows that produce more milk increases. For example, Strand- berg (1996) estimated that an increase in longevity from three to four lactations increases average milk yield per

Upload: gregor-gorjanc

Post on 28-Nov-2014

63 views

Category:

Documents


6 download

DESCRIPTION

Potočnik, K., Gantner, V., Krsnik, J., Štepec, M., Logar, B., Gorjanc, G. 2011. Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica, 98(2): 93-100.

TRANSCRIPT

Page 1: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta argiculturae Slovenica, 98/2, 93–100, Ljubljana 2011

doi:10.2478/v10014-011-0025-5 COBISS:1.01Agriscategorycode:L01

ANALYSISOFLONGEVITYINSLOVENIANHOLSTEINCATTLE

KlemenPOTOČNIK1,VesnaGANTNER2,JurijKRSNIK1,MiranŠTEPEC1,BetkaLOGAR3,GregorGORJANC1

ReceivedAugust20,2011;acceptedSeptember22,2011.Delojeprispelo20.avgusta2011,sprejeto22.septembra2011.

1 Univ.ofLjubljana,BiotechnicalFac.Dept.ofAnimalScience,Groblje3,SI-1230Domžale,Slovenia2 J.J.StrossmayerUniv.inOsijek,Fac.ofAgriculture,TrgSvetogTrojstva3,31000Osijek,Croatia3 AgriculturalinstituteofSlovenia,Hacquetova17,SI-1000Ljubljana,Slovenia

Analysis of longevity in Slovenian holstein cattleThelongevityofSlovenianHolsteinpopulationwasana-

lysedusingsurvivalanalysiswithaWeibullproportionalhaz-ardmodel.DataspannedtheperiodbetweenJanuary1991andJanuary2010for116,200cowsfrom3,891herds.Longevitywasdescribedas the lengthofproductive life– fromfirstcalvingtillcullingorcensoring.Recordsabovethesixthlactationwerecensored to partially avoid preferential treatment. Statisticalmodelincludedtheeffectofageatfirstcalving,stageoflacta-tionwithinparity,yearlyherdsizedeviation,seasondefinedasyear,herd,andsire-maternalgrandsire(mgs).Someeffectshadtimevaryingcovariates,whichleadto1,839,307oronaverage16elementaryrecordspercow.Herdandsire-maternalgrand-sireeffectsweremodelledhierarchically.Pedigreeforsiresandmaternalgrandsiresincluded2,284entries.Estimatedvariancebetweenherdswas0.12,whilebetweensirevariancewas0.04.Heritabilitywasevaluatedat0.14.Genetictrendforsireswasunfavourable,butnotsignificant.Afurtherresearchisneededtodefinetherequirednumberofdaughterspersireandthedy-namicsofgeneticevaluationforsireswhosemajorityofdaugh-tersstillhavecensoredrecords.

Key words:cattle/breeds/SlovenianHolstein/longevity/Weibullproportionalhazardsmodel

Analiza dolgoživosti pri črno-beli pasmi goveda v SlovenijiZa analizo dolgoživosti smo pri slovenski črno-beli po-

pulaciji govedi uporabili metodologijo analize preživetja inWeibullov model sorazmernih ogroženosti. V analizo smovključilipodatke116.200kraviz3.891čredskoziobdobjeodjanuarja1991dojanuarja2010.Dolgoživostjebilapredstavlje-nakotdobaproduktivnegaživljenja,kijedefiniranakotšteviloodprvetelitvedoizločitvealidodatumazajemapodatkovzaživali,kisonatadatumbilešežive.Šestoinkasnejšelaktacijesmookrnilinakonecšeste laktacije,dasmoomililiprecenje-nostboljšihživali.Vstatističnimodelsmovključilivplivsta-rostiobprvitelitvi,stadijalaktacijeločenozavsakozaporednolaktacijo,spreminjanjevelikostičredemedleti,letotelitve,čre-do,očeta inmaterinegaočeta.Ravninekaterihvplivovsoča-sovnospremenljivi,karpovzroči,dasmovanaliziobravnavali1.839.307 zapisov ali povprečno 16 osnovnih zapisov na kra-vo.Čredainvplivočetazmaterinimočetomstabilavmodelvključenahierarhično.Rodovnikzaočeteinmaterineočetejeobsegal2.284zapisov.Ocenjenavariancazavplivčredejezna-šala0,12,medtemkojeocenavariancemedočetiznašala0,04.Dednostnideležjebilocenjenna0,14.Genetskitrendimane-gativnosmer,anistatističnoznačilen.Potrebnebodonadaljnjeraziskave,dabomodoločilizadostnoštevilohčerapobikuindinamiko obračunov plemenskih vrednosti za bike, ki imajovečinohčeraševfaziprireje.

Ključne besede:govedo/pasme/slovenskačrno-belapa-sma/dolgoživost/Weibullovmodelsorazmernihogroženosti

1 INTRODUCTION

Longevityisatraitwithgreatimpactondairypro-ductioneconomyand is, therefore,ofconsiderable im-portanceindairycattlebreedingprogrammes(Charffed-

dineet al.,1996;StrandbergandSoelkner,1996).Withtheincreaseoflongevity,theproportionofmaturecowsthatproducemoremilkincreases.Forexample,Strand-berg(1996)estimatedthatanincreaseinlongevityfromthreetofourlactationsincreasesaveragemilkyieldper

Page 2: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta agriculturae Slovenica, 98/2 – 201194

K.POTOČNIKet al.

lactationandprofitperyearbetween11and13%.Inad-dition,improvementinlongevitydecreasesreplacementcostsandsomewhatincreasesselectionintensity.

There are several ways to implement selection onlongevityinthebreedinggoal,directlyorindirectly.Di-rect longevitycanberepresentedasthe lengthof(pro-ductive) life(LPL)orstayability.IncattlebreedingLPLis usually defined as the elapsed time between the firstcalvingandculling,while stayability isdefinedasabi-narytrait thatmeasurescowsurvival(liveorculled)atacertainpointintime.TheuseofLPLispreferredsincestayability as a discrete trait provides less information.Unfortunately, LPL, as well as stayability, can be quan-tified only after the cows are culled, though both ap-proachesprovidepartialinformationwhencowsurvivesto the next “period” in life. Therefore, the informationonthelongevityofdaughtersofasirebecomesavailablewiththeincreasingageofasire.Thisinherentlyleadstothe prolonged generation interval. Low heritability forlongevity(ShortandLawlor,1992;VollemaandGroen,1996) induces unreliable estimation of breeding values(BV)basedonlyontheinformationofparentsorgrand-parents.

Due to long generation interval, breeding pro-grammesalsoincludeindirectmeasuresoflongevityviacorrelatedtraitssuchas fertility,health,andconforma-tiontraits(Burnsideet al.,1984).Additionalgainisdueto the fact that the data on these indirect traits can becollectedrelativelyearlyinthelifeofacow.Nonetheless,both representations of longevity (direct and indirect)haveameritinamodernbreedinggoal(Essl,1998).

Analysis of indirect representations of longevityis to a large extent done with a standard linear modelbased on the Gaussian (normal) distribution. SpecificapproachisneededforaproperanalysisoftheLPL,dueto the presence of live animals at the time of analysis(censored records) and changes in culling criteria overtheproductivelifeofcows(timevaryingcovariates)(e.g.Ducrocq et  al., 1988a). Exclusion of censored recordsfromtheanalysis,ortreatingthemasuncensoredleadstobiasedresults(Ducrocq,1994).Additionally,relation-shipbetweenlongevityanditseffectsisrathermultipli-cativethanadditive(e.g.Ducrocqet al.,1988a).Survivalanalysis can handle this kind of data. In the last yearsseveralcountriesintroduceddirectlongevityintherou-tine genetic evaluation of cattle and most of them usethe Weibull proportional hazard model (INTERBULL,2009),whichrepresentsaclassofmodelsinthefieldofsurvival analysis. Other statistical approaches (models)can also be used, but proportional hazard model havebetter properties (e.g. Caraviello et  al., 2004; Jamroziket al.,2008;Potočniket al.,2008).

Theaimof thisstudywas topresent theresultsof

genetic evaluation for the length of productive life inSlovenian Holstein population using a Weibull propor-tionalhazardmodel.

2 MATERIAL AND METHODS

2.1 DATA

Rawdatafor126,716SlovenianHolsteincowsbornfrom1982to2008wereprovidedbytheAgriculturalIn-stituteofSlovenia.Inordertouseolddatabuttoavoidmodelling thedataup to theyear1991, the truncationdatewassetatJanuary1st1991.Ontheotherside,thedateof last data collection was January 29th 2010. For cowsaliveatthattimelongevitywastreatedasrightcensored.Longevity was defined as the length of productive life(LPL)andwascalculatedasthenumberofdaysfromthefirstcalvingtoculling(uncensored/completerecords)orto themomentofdatacollection(incomplete/censoredrecords). The LPL of cows surviving beyond the sixthlactationwasalsocensored inorder toavoid theeffectofpreferentialtreatmentandtofocusonearlycullinginthelifeofacow.Cowswithmissingorinconsistentdatawithin the defined limits were removed (29,252 cows):culling before the date of truncation, calving date afterthedateofculling,noinformationfor600 daysaftercalv-ing,missingdataforthefirstthreelactations,daughtersofsireswithlessthan20daughters,andmissingcovariateorfactordata.

Thestructureofuseddataanddescriptivestatisticsare given in Table  1. Altogether LPL for 116,200 cowsfrom3891herdswereusedintheanalysis.Cowsintheanalysis were daughters of 707 sires, while the wholesire-maternalgrandsirepedigreeconsistedof2,284sires.Cowswereonaverageculledinthethirdlactation,which

Cows,no. 116,200Sires,no. 707Pedigree,no. 2,284Censoredrecords,% 41.0Numberoflactationsinlife

uncensoredrecords 3.0censoredrecords 3.0

Lengthofproductivelife,daysuncensoredrecords 1,095±660censoredrecords 1,129±754

Table 1: Structure of data and descriptive statistics (± standard deviation)Preglednica 1: Struktura podatkov in opisna statistika (± stan-dardni odklon)

Page 3: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta agriculturae Slovenica, 98/2 – 2011 95

ANALYSISOFLONGEVITYINSLOVENIANHOLSTEINCATTLE

amountedto1,095 daysofproductivelife.Percentageofcensoredrecordswas41.0%.Censoredrecordshadaboutthesamemeans,butlargervariability.

2.2 SURVIVALANALYSIS

Weibull proportional hazards model was used fortheanalysisofLPL.ThismodelisbuiltupontheWeibulldistribution,whosedensity(1)andhazard(2)functionforthei-threcordtiare:

(ti|λ,ρ)=λρ(λti)ρ−1 exp(−(λti)

ρ), (1)

h(ti |λ,ρ)=λρ(λti)ρ, (2)

whereλ(scale)andρ(shape)arestrictlypositiveparam-eters. In proportional hazard model it is assumed thatthebaselinehazardfunctionchangesproportionallywithchangeincovariate(s)orfactorlevels.FortheanalysisofLPLthehazardfunctionwasmodelledas:

h(tijklmnop |λ,ρ,else)=h0(tijklmnop |λ,ρ)exp(ci+lj+yk+hl+dm+sn+1/2so), (3)

where:h(tijklmnop |λ,ρ,else)=hazardof cullingp-th cowgivenotherpa-

rameters,h0(tijklmnop |λ,ρ) =baselineWeibullhazardfunction(2),ci = i-th age at first calving: 0 (unknown) and

from19to50 months,lj = j-thlactationstage(1–60 days,61–150 days,

151–270 days,271-daystilldrying,anddryperiod)withinparity–altogether30levels;timevaryingfactor,

yk =k-th season defined as year (1990–2010);timevaryingfactor,

hl = l-thherd(3891levels);timevaryingfactor,dm =m-thherd sizedeviation incomparison to

previousyear(≤-70%,(-70%,-40%],(-40%,-10%], (-10%, 10%], (10%, 40%], (40%,70%],and>70%);timevaryingfactor,

sn+1/2so =n-th sire and the o-th maternal grandsire(onwardsbotheffectsaretermedsireeffect)ofthep-thcow.

Levelsoftimevaryingfactors(lactationstagewith-in parity, year, herd, and herd size deviation) changedwith cow “status” changes in time creating subsequentelementary records, while levels for others effects wereconstantoverwholelifetimeofacow.Altogether,therewere1,839,307elementaryrecords.Herdandsireeffectswere modelled hierarchically: log-gamma distributionfor herd effect and multivariate normal for sire effectwith additive genetic covariance matrix build from the

pedigree.TheusedWeibullproportionalhazardsmodeland thecorrespondingassumptionscanbe sketched inmatrixformas:

y|b,h,s,ρ ~ Weibull (Xb+Zh+Ws, ρ), (4)

h|γ ~ Log − Gamma (γ,γ), (5)

s|G ~ Normal (0,G), (6)

where:

b=avectorwithinterceptρ ln(γ))andparametersforthefollowingeffects: ageatfirst calving, stageof lactationwithinparity, year,andthedeviationofherdsizefromyeartoyear,

h= thevectorofparametersforherdeffect,s = thevectorofparametersforsireeffect,γ=Log-Gammadistributionparameter,G=additivegeneticcovariancematrixamongsires–aproductofnu-

meratorrelationshipmatrixbetweensiresAsandadditivegeneticvariancebetweensires(σs

2).

Heritabilityaccordingtothemodel(3,4–6)wascal-culatedfollowingMeszaroset al.(2010):

, (7)

where:

σs2+¼σs

2 isvarianceduetosireandmaternalgrandsireeffects(3),

)(log)( 2

2)1( x

xx Γ

∂∂

=ψisatrigammafunctionusedtoevaluatethevari-

anceoflog-gammaprocess(5)givingbetweenherdvariance(σh2),while

thevalueof1istheunderlyingresidualvariance.

DataprocessingwasdonewithSASsoftwarepack-age(SASInstitute,2000),whileSurvivalKitversion3.10(Ducrocq and Soelkner, 1998) was used for modellingandparameterestimation.Inthefirststepaseriesoflog-likelihoodratiotestswereperformedforeffectsthatwerenotmodelledhierarchically–theimportanceofeachef-fectwastestedasacomparisonbetweenthefullmodelandthemodelwheretheeffectundertestingwasexclud-ed.Inthenextstepherdandsireeffectswereaddedtothemodeltoobtainestimatesforallmodelparameters.Inresultsrelativeriskisequaltothevalueofsolutionsformodelparametersonexponentialscale(3)proportionaltosomespecifiedbaselinelevelthathasasolutionequalto1(e.g.26 monthsfortheageatfirstcalving).Eachplotof relative risks is also augmented with the number ofcensored and uncensored records per level of a factor.Inthecaseoftimevaryingfactorsonlythelastelemen-taryrecordofacowwasconsideredforcomputingthenumberofrecordsperlevelofafactor.

Page 4: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta agriculturae Slovenica, 98/2 – 201196

K.POTOČNIKet al.

3 RESULTS AND DISCUSSION

All effects included in the model were highly sig-nificant (P  <  0.001), which is not surprising given thesize of data set and the previous knowledge of effectimportance forLPL.Distributionofageatfirst calvingwasexpectedwiththemajorityofcowsintherangebe-tweentwoandthreeyearsofage(Figure 1).Relativerisk

of culling increasedalmost linearlywith the increasingage at first calving. Similar results were obtained alsobyVollemaandGroen(1998)andRogerset al. (1991),whileothersdidnotfoundsignificance(Ducrocqet al.,1988a;Ducrocq,1994)orconcludedthatthiseffectwasnotimportant(Vukasinovicet al.,1997).Thiscanbeatleastpartiallyattributedtothefactthatourresultsdonotdirectly implycausal relationshipbetweenLPLand the

Figure 1: Relative risk of culling and number of records by age at first calvingSlika 1: Relativno tveganje za izločitev in število zapisov glede na starost ob prvi telitvi

Figure 2: Hazard of culling and number of records by stage of lactation within paritySlika 2: Ogroženost za izločitev in število zapisov glede na stadij laktacije in zaporedno laktacijo

Page 5: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta agriculturae Slovenica, 98/2 – 2011 97

ANALYSISOFLONGEVITYINSLOVENIANHOLSTEINCATTLE

age at first calving. Results only imply that there is as-sociationbetweenLPLandtheageatfirstcalvinginourpopulation,whichindicatesthatcowsthathadlatefirstcalvinghadalsosomeotherproblems(likelyrelated toreproductivesuccess)thatincreasedriskofbeingculledearly.Estimatesatthestartandtheendofconsideredageinterval were very variable due to the smaller numberofrecords.Givenalmostlinearrelationship,variablere-sultsatmargins,andthatageistimeindependenteffectapossibleapproachwouldbe tomodel thiseffectwithlinearregression.Regressionisnotappropriatefortimedependenteffectsduetotheexplosionofnumberofel-ementaryrecords.

Thestageofproductionhasa significanteffectonrisk of a cow being culled due to biological (increasedprobabilityofmastitisat thestartof lactation)or tech-

nological factors (owners’decisions in thedryperiod).Since the stage of lactation within parity changes withincreasingage(dependentvariable)werepresentedthiseffectusingbaselinehazardfunction(3)multipliedwiththecorrespondingrelativeriskforeachstagewithinpar-ity (Figure  2) for a fixed calving interval of 400  days.Numberofculledcowswashighestinthesecondparityanddroppedinlaterparities.Hazardincreasedovertimewithconsiderablechangesattheendoflactation–thatisintheperiodbetween271 daysafterlactationanddry-offandinthedryperiod.Virtuallythesameresultswereobtainedalsoinotherstudies(e.g.Ducrocq,1994;Vuka-sinovicet al.,1997;Potočniket al.,2010).Thefirstparityshoweduniquepatternwithincreasedhazardinthefirsttwoperiodsoflactation(1–60and61–150 days),whichmightbeduetothehigherincidenceofhealthdisordersduring early lactation. Similar estimates were obtainedalsobyDucrocq(1994)andVukasinovicet al.(1997).

Herd size dynamics has also influence on cullingcriteria. In general herd expansion lowers risk, whilerisk ishigher inshrinkingherds.Herds inSloveniaareingeneralsmall,sothereissubstantialvariabilityinherdsizechangesfromyeartoyear.Majorityofrecords(cen-soredornot)wereintherangeof−40to40%ofherdsizechange.Relativeriskforcullingwasratherconstantforherdsizechangelevelsabove−40%,whileitincreasedforthetwolevelsbellowthisthresholdasexpected–cowsfromherdswithdecreasingsizehave largerprobabilityofbeingculled.Weigelet al.(2003)calculatedtherela-tive culling risk of high producing (top 20%) and low

Hyperparameter/Quantity EstimateShape,ρ 2.00Log-gammaparameter,γ 8.50Betweenherdvariance,σh

2=ψ(1)(γ) 0.12Betweensirevariance,σs

2 0.04Additivegeneticvariance,σa

2=4σs2 0.16

Heritability, 0.14

Table 2: Estimates of model hyperparameters and derived quantitiesPreglednica 2: Ocene parametrov modela in izpeljanih količin

Figure 3: Relative risk of culling and number of records by levels of variation in herd size between yearsSlika 3: Relativno tveganje za izločitev in število zapisov glede na letno spremembo velikosti črede

Page 6: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta agriculturae Slovenica, 98/2 – 201198

K.POTOČNIKet al.

producing (bottom 20%) cows relative to average cowsinthesameherdwithregardtoherdsizechanges.Theydeterminedthat,beforeherdexpansion, lowproducingcowswere4.2timesmorelikelytobeculledthanaveragecows,whilehighproducingcowswereonly0.5timesaslikelytobeculledasaveragecows.Afterherdexpansion,therelativeriskforlowproducingcowsdroppedto2.6timesthatofaveragecows,andtheriskforhighproduc-ing cows increased slightly to 0.7 times that of averagecows,whichclearlyshowstheeffectofherdexpansiononthereducedriskofculling.

Cullingcriteriaalsochangewithtime.Possiblerea-sonsarediseaseoutbreakinpopulation,changeofprices,milkquota,etc.Inordertocapturesuchvariationswein-cludedinthemodeleffecttoseasondefinedasyear.Ourdataspannedperiodbetween1991and2010(Figure 4).Relativeriskofcullingwasverylowinthefirstyearsduetolackofrecordsinthatperiod,butreachedoveralllevelaftertheyear1995.Afterthisyearweobserveoverallin-creaseinrelativeriskofcullingoveryears.Asharpde-creaseinriskwasestimatedfortheyear2002,whichcanbeattributedtofarmers’decisiontokeepmoreanimalsonfarminordertogethighermilkquotaandsubsidiesper farm with the forthcoming new quota and subsidysysteminSloveniaatthatperiod.Riskwaslogicallyverylowinthelastyear(2010)duetothelargenumberoflive(censored)animalsintheanalysis.

Based on the used statistical model, between sirevariancewasestimatedto0.04,whilebetweenherdvari-ancewasestimatedto0.12(Table 2).Thesevaluesareonexponential scale of the Weibull model (3) and do not

have meaningful units related to the analysed variable(LPL).EstimatedheritabilityusingtheapproachofMes-zaroset al.(2010)was0.14,whichissimilartotheval-uesreportedinAustria(0.18)andGermany(0.16)andrelativelyhighcomparingwithothercountries thataremembersofINTERBULL(INTERBULL,2009).

Breeding values for LPL were presented on scalewithmean100and standarddeviation12with favour-able values (longer LPL) above 100. Genetic trend byyearofbirthforsiresintheperiod1984–2005showsthatthere was no selection on longevity (Figure  5). Overalltrend was negative (−0.12 ± 0.14), but not significant(p=0.385).Differencesbetweenyearswereminimalex-cept for the last four years. This could be attributed tothesmallernumberofevaluatedsiresandtothefactthatthesesireshavealotofdaughterswithcensoredrecords.

Theaccuracyofgeneticevaluationhighlydependsontheratioofcensoredanduncensoredrecords.Astheproportionofcensoredrecordsdecreases,theevaluationaccuracyincreases.Also,itisnecessarytohavesufficientnumberofdaughterspersire.Vukasinovicet al.(1997)stated that more than 30 to 40% of censored recordswould lead to inaccurate results. Same authors statedthatsmallnumberofdaughterspersirewithoutanyoronly few uncensored records biases sires ranking. Egg-er-Danner et  al. (1993) performed retrospective studywheretheycomparedtherankingofsiresfromafulldatafilewithoutcensoredrecordsandfromatruncateddatawith a different proportion of censored records. Theyobservedthatrankcorrelationsbetweenbreedingvaluesdroppedastheproportionofcensoringincreased.Fur-

Figure 4: Relative risk of culling and number of records by yearSlika 4: Relativno tveganje za izločitev in število zapisov glede na leto

Page 7: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta agriculturae Slovenica, 98/2 – 2011 99

ANALYSISOFLONGEVITYINSLOVENIANHOLSTEINCATTLE

therresearchisneededinourpopulationtodeterminetheimpactofcensoredrecordsontheaccuracyofgeneticevaluationaswellastodeterminehowmanydaughterspersireareneeded.

4 CONCLUSION

Survivalanalysismethodologywasapplied for thegeneticevaluationoflongevity(definedasthelengthofproductive life) in Slovenian Holstein cattle. Statisticalmodelincludedtheeffectofageatfirstcalving,lactationstagewithinparity,yearlyherdsizedeviation,year,herd,andadditivegenetic effect as capturedby sire andma-ternalgrandsireeffects.Parameterestimatesweresimilartostudiesinothercountries.Genetictrendwasslightlynegative (unfavourable), yet not significant. Relativelyhigh differences between average breeding values wereobservedinlastyears.Asaccuracyofgeneticevaluationhighlydependson thenumberofdaughtersperevalu-ated sire and on the ratio of censored and uncensoredrecordsfurtherinvestigationsareneeded.

5 REFERENCES

Burnside E.B., McClintock A.E., Hammond K. 1984. Type,productionandlongevityindairycattle:areview.AnimalBreedingAbstracts,52,711–719

Caraviello D.Z., Weigel K.A., Gianola D. 2004. ComparisonbetweenaWeibullproportionalhazardsmodelanda lin-

ear model for predicting the genetic merit of US Jerseysires fordaughter longevity. JournalofDairyScience,87,1469–1476.http://jds.fass.org/cgi/content/abstract/87/5/1469(26.Mar.2010)

CharffeddineN.,AlendaR.,CarabanoM.J.,BejarF.1996.Se-lectionfortotalmeritintheSpanishHolstein-Friesian.IN-TERBULLBulletin,12,142–146

DucrocqV.1994.Statisticalanalysisoflengthofproductivelifefor dairy cows of the Normande breed. Journal of DairyScience, 77, 855–866. http://jds.fass.org/cgi/content/ab-stract/77/3/855(26.Mar.2010)

DucrocqV.,QuaasL.R.,PollakE.J.,CasellaG.1988a.Lengthofproductivelifeofdairycows.1.JustificationofaWeibullmodel.JournalofDairyScience,71,3061–3070.http://jds.fass.org/cgi/content/abstract/71/11/3061(26.Mar.2010)

DucrocqV.,QuaasL.R.,PollakE.J.,CasellaG.1988b.Lengthofproductivelifeofdairycows.2.Variancecomponentes-timationandsireevaluation.JournalofDairyScience71,3071–3079.http://jds.fass.org/cgi/content/abstract/71/11/3071(26.Mar.2010)

DucrocqV.,SoelknerJ.1998.TheSurvivalKit–V3.0:APack-ageforLargeAnalysisofSurvivalData.In:Proceedingsofthe6thWorldCongressonGeneticsAppliedtoLivestockProduction,Armidale,Australia,22,51–52

Egger-Danner C., Soelkner J., Essl A. 1993. Zuchtwertschät-zung für Merkmale der Langlebigkeit: Vergleich der Pro-portionalHazardsAnalysefürdasAusfallsrisikomitBLUPfürdieVerbleiberate.VortragstagungderDGfZ/GfT,Göt-tingen,Germany

EsslA.1998.Longevityindairycattlebreeding:areview.Live-stockProductionScience,57,79–89.http://dx.doi.org/10.1016/S0301-6226(98)00160-2(26.Mar.2010)

Figure 5: Genetic trend for the length of productive life in Slovenian Holstein siresSlika 5: Genetski trend za dolgoživost pri plemenjakih slovenske črno-bele pasme

Page 8: Analysis of longevity in Slovenian Holstein cattle. Acta Agriculturae Slovenica

Acta agriculturae Slovenica, 98/2 – 2011100

K.POTOČNIKet al.

INTERBULL 2009. Interbull Routine Genetic Evaluation forDirectLongevityJanuary2009–Table3.http://www-interbull.slu.se/longevity/l-table3-091.html(20.Jan.2009)

JamrozikJ.,FatehiJ.,SchaefferL.R.2008.Comparisonofmod-elsforgeneticevaluationofsurvivaltraitsindairycattle:asimulationstudy.JournalofAnimalBreedingandGenetics,125,75–83.http://dx.doi.org/10.1111/j.1439-0388.2007.00712.x(26.Mar.2010)

MeszarosG.,PalosJ.,DucrocqV.,SoelknerJ.2002.HeritabilityoflongevityinLargeWhiteandLandracesowsusingcon-tinuoustimeandgroupeddatamodels.GeneticsSelectionandEvolution,42,13.http://dx.doi.org/10.1186/1297-9686-42-13(26.May2010)

PotočnikK.,KrsnikJ.,ŠtepecM.,GorjancG.2008.Compari-sonbetweenmethodsforestimationofbreedingvaluesforlongevity in Slovenian Holstein population. INTERBULLBulletin,38,61–65

PotočnikK.,Gorjanc,G.,Štepec,M.,KrsnikJ.2010)Geneticevaluation for longevity in Slovenian Simmental cattle.ActaAgrariaKaposváriensis,14,153–160

RogersG.W.,HargroveG.L.,CooperJ.B.,BartonE.1991.Rela-tionshipsamongsurvivalandlineartypetraits inJerseys.JournalofDairyScience,74,286–291.http://jds.fass.org/cgi/content/abstract/74/1/286(26.Mar.2010)

SASInstitute.2000.TheSASSystem.Version8.Cary,SASIn-stitute

ShortT.H.,LawlorT.L.1992.Geneticparametersofconforma-tiontraits,milkyield,andherdlifeinHolsteins.JournalofDairyScience,75,1987–1998.

http://jds.fass.org/cgi/content/abstract/75/7/1987(26.Mar.2010)

StrandbergE.1996.Breeding for longevity indairy cows. In:ProgressinDairyScience.Phillips,C.J.C.(ed.).Oxon,CABInternational,Wallingford,125–144

Strandberg E., Soelkner J. 1996. Breeding for longevity andsurivalindairycattle.In:ProceedingsoftheInternationalWorkshoponGeneticImprovementofFunctionalTraitsinCattle.Gembloux,Belgium,12,111–119

VollemaA.R.,GroenA.F.1996.Geneticparametersoflongev-itytraitsofanupgradingpopulationofdairycattle.JournalofDairyScience,79,2261–2267.http://jds.fass.org/cgi/content/abstract/79/12/2261(26.Mar.2010)

Vollema A.R., Groen A.F. 1998. A Comparison of BreedingValuePredictors forLongevityUsingaLinearModelandSurvivalAnalysis.JournalofDairyScience,81,3315–3320.http://jds.fass.org/cgi/content/abstract/81/12/3315(26.Mar.2010)

VukasinovicN.,Moll J.,KuenziN.1997.AnalysisofProduc-tive Life in Swiss Brown Cattle. Journal of Dairy Science80,2372–2579.http://jds.fass.org/cgi/content/abstract/80/10/2572(26.Mar.2010)

WeigelK.A.,PalmerR.W.,CaravielloD.Z.2002.Investigationoffactorsaffectingvoluntaryandinvoluntarycullinginex-pandingdairyherdsinWisconsinusingsurvivalanalysis.JournalofDairyScience,86,1482–1486.http://jds.fass.org/cgi/content/abstract/86/4/1482(26.Mar.2010)