Relationships between estimated breeding values for claw health and production as well as functional traits in dairy cattle

Transcription

1 Introduction Relationships between estimated breeding values for claw health and production as well as functional traits in dairy cattle H. Alkhoder 1), R. Pijl 2), and H. H. Swalve 1) 1) Institute of Agricultural and Nutritional Sciences, Theodor-Lieser-Str. 11, Halle/Saale, Germany 2) Fischershäuser 1, Jever, Germany Claw diseases are a major issue in dairy cows Culling rate is over 12 % per year for claw disorder Costs per lameness are between The genetic improvement of claw health is possible Genetic correlations between milk yield and claw disorder were in the range of 0.06 to 0.34 (KÖNIG et al. 2005) antagonistic 61 st Annual Meeting of the European Association for Animal production, August 23 rd 27 th, 2010 HERAKLION CRETE ISLAND, GREECE, Session 09, Paper 02, Book of Abstracts p 67 Thedatabase Records taken at hoof trimming (whole herd) by the professional hoof trimmer René Pijl during 2002 to 2010 Diagnoses (Laminitis, Dermatitis digitalis, etc.) Plus: Pedigree data from VIT, Verden Data on milk yield, fertility, longevity from VIT René Pijl PDA Since the data contains new clients as well as old ones, different subsets of data can be extracted for different types of analysis herds visited up to 17 times 166 claw trimming dates 929 cows up to 17 times treated complete records Hatem Alkhoder Taking records at trimming Laminitis

2 Incidence rates of disorders in four subsets of the data Rotation Dermatitis Digitalis Laminitis All data Heifers at first visit Disorder status per Lifetime disorder status (If 4 observations per cow) n=8,106 % Disorder n=79,181 n=16,287 n=40,033 % % % Laminitis LAM Dermatitis Digitalis DD Dermatitis Interdigitalis DID White line disease WLD Sole ulcer SU Rotation ROT Tylom (Hyperplasia) TYL Thick hock TH Dermatitis interdigitalis White line disease Sole ulcer This includes clinical as well as sub-clinical cases! Statistical models PR (Y hjklm =1) = θ (µ +CACL(LN) hk + BB j + LN k + DIMCL(LN) lk + PU m ) ** PR (Y hjklm =1) = θ (µ +CACL(LN) hk + BB j + LN k + DIMCL(LN) lk + AN m + PU m ) * Y hjklmn = µ +CACL(LN) hk + BB j + LN k + DIMCL(LN) lk + AN m + PU m + e hijklmn * Y hjklm, Y hjklmn = trait of claw disorder (1 = positive 0 = negative) PR =, θ = link function (logit and probit), µ = overall mean CACL(LN) hk = fixed effect of calving age h in classes (1,..., 4) nested within number k BB j = fixed effect of herd-visit j (1,..., 918) LN k = fixed effect of number k (k = 1, 2,, 6) DIMCL(LN) lk = fixed effect of days in milk l (l = 1,, 8: with classes of 50 days) nested within number k AN m = random animal effect for animal m (additive genetic variance) PU m = random permanent environmental effect of animal m e hjklmn = random rest effect * ASREML 2.0 ** proc glimmix SAS 9.1

3 LSMEANS of incidence rate for parity (1) LSMEANS of incidence rate for parity (2) LAM PATH disorder LAM PATH LAM = Laminitis DD = Dermatitis Digitalis DID = Dermatitis Interdigitalis WLD = White Line Disease SU = Sole ulcer TH = Thick hocks TYL = Tylom (Hyperplasia interdigitalis) PATH = any disease of the above ROT = Rotation DD ROT disorder DD ROT disorder DSG TH KSG SU TYL disorder DSG TH KSG SU TYL LAM = Laminitis DD = Dermatitis Digitalis DID = Dermatitis Interdigitalis WLD = White Line Disease SU = Sole ulcer TH = Thick hocks TYL = Tylom (Hyperplasia interdigitalis) PATH = any disease of the above ROT = Rotation DID WLD disorder DID WLD disorder day in milk day in milk all lac. 5 6 all lac age in class age in class Laminitis: LSMEANS of incidence rate for stage of and age at calving Incidence rates for disorders in the respective best and worst herds Herd LAM DD DID WLD SU Observations N Visits N Ave. milk yield (kg) A b B w C b D w E b F w G b H w I b J w b = best herd w = worst herd

4 Estimates of heritabilities and proportion of permanent environmental effect in three subsets of the data Heritability estimates Ratio of permanent environ. effect All parities Parity1 Par. 2 All parities Par. 2 Linear Logit Probit Logit Logit Linear Logit Probit Logit LAM DD DID WLD SU ROT TYL TH SE Estimates of heritabilities; genetic correlations (below diagonal) LAM DD DID WLD SU ROT TYL TH LAM (0.009) DD - (0.061) DID 0.22 (1) WLD 0.33 (7) SU 0.48 (5) ROT (9) TYL 0.29 (0.067) TH (0.063) (0.072) 0.01 (0.078) 0.09 (0.071) 0.04 (0.083) - (0.13) (0.065) (0.069) (0.12) (0.073) - (0.048) - (0.065) (0.077) 0.34 (0.13) (0.026) (8) () (0.019) (0.007) Estimates of heritabilities; genetic correlations (below diagonal) in comparison with correlations between EBV for sires with more than four daughters and an accuracy 80 % (above diagonal) LAM DD DID WLD SU ROT TYL TH LAM * * * - * (0.009) DD - (0.061) DID 0.22 (1) WLD 0.33 (7) SU 0.48 (5) ROT (9) TYL 0.29 (0.067) TH (0.063) (0,072) 0.01 (0.078) 0.09 (0.071) 0.04 (0.083) - (0.13) 0.38 * * * (0.065) (0.069) (0.12) * * * (0.073) - (0.048) - (0.065) (0.077) 0.34 (0.13) = number of sires * = significant * (0.026) (8) () * (0.019 ) (0.007) Number of sons within quartiles of high / low EBV for claw disease for five sires of sons (Ns = number of sons per sire-of-sons, No = number of observations, Ne = number of grand dghts) trait quartile S Ns=12 Ne=322 No=1058 A Ns=17 Ne=615 No=1058 P Ns=15 Ne=628 No=2201 B Ns=28 Ne=1193 No=4431 L Ns=14 Ne=871 No=3177 LAM High 7 * 11 * 12 * 16 * 2 Low * DD High * 3 13 * Low 7 * 10 * 1 23 *. DID High * 11 5 Low 7 * 8 * WLD High Low 4 9 * 11 * * SU High 9 * 8 * 15 * 17 * 1 Low * High = High resistance / low incidence Low = Low resistance / high incidence * indicates the quartile of the sire

5 Genetic correlations between official EBV for feet & leg traits and own EBV for resistance to claw disorder for two subsets of sires (Gr. 1 accuracy 70% and Gr. 2 accuracy 85% ) Genetic correlations between official EBV for trait complexes and own EBV for resistance to claw disorders (accuracy 0.85 ) Trait Gr. LAM DD DID WLD SU ROT TYL TH Locomotion * * 0.11 * * * * Feet & Legs * * * 0.12 (Final Score) 2 * * Rear-legsrear-view * * * Rear-legsside-view * Foot angle * * * * * Hock quality * 0.19 * * * Number of sires König and Swalve (2006); model calculation: Selection for conformation is not enough! LAM DD DID WLD SU ROT TYL TH Conformation 0.22 * Total Merit 0.37 * * Dairy production Longevity 0.51 * * * Fertility 0.34 * Udder health Number of sires * = significant * = significant Strategies to estimate the relationship between milk production traits and claw disorders S1* Correlations between own sire EBV for claw disorders with national EBV for production traits S2* Own sire EBV for production from RRM and own sire EBV for claw disorders; 1 st s only S3* Direct estimation of genetic correlations between claw disorders and 305-d production; 1 st s only S4* Direct estimation of genetic correlations between claw disorders and milk at time of trimming; first three s S5** Estimation of the regression of claw disorders status on 305-d milk yield y ijkl = HTD i + Σ c jm.f jm + Σ b kn.a kn + Σ b kn.p kn + e ijkl y ijkl HTD i f jm c jm a kn and p kn b kn Model for EBV for production traits e ijlkl = milk production trait on test day n in the first for animal k = fixed effect of Herd-Test-Day I (1,,10370) = regression coefficient m of fixed curve for group of calving year-calving season-calving interval-calving age j (1,..,472) = m-th term of Ali & Schaeffer function of fixed curve = m-th random regression coefficient of animal k for genetic effect(1,.., 64900) and permanent environmental effect (1,..,17295) = m-th term of Legendre polynomial function (order 2) of random curve for animal k = random residual error (1,,156987) S2 Random Regression Animal Model Legendre Polynomials (order 2) * ASREML 2.0 ** proc glimmix SAS 9.1

6 Multivariate sire model for milk yield and disorders -1 st s only PR (Y ijkm =1) = θ (µ +CACL i + BB j + DIMCL k + S m ) M ijklmn = µ + CACL i +BET j + CICL k + YS l + S m + e ijklmn Y ijkm = trait of claw disorder (1 = positive 0 = negative) PR =, θ = link function (logit and probit), µ = overall mean CACL i = fixed effect of calving age i in classes (1,..., 4) BB j = fixed effect of herd-visit j (1,..., 528) DIMCL l = fixed effect of days in milk l (l = 1,, 8: with classes of 50 days) M ijklmn = trait of milk ( production) yield in the first BET j = fixed effect of herd j (1,, 109) Ys l = fixed effect of year-season I (1,,35) CICL k = fixed effect of calving interval k in 5 classes S m = random sire effect for sire m (1,,425) e ijklmn = random residual effect (1,,10486) Multivariate sire model S3 Threshold model for disorders Linear model for production milk yield 1 st s only Bivariate animal model for milk yield at trimming and disorders - First three s PR (Y hjklm =1) = θ (µ +CACL(LN) hk + BB j + LN k + DIMCL(LN) lk + AN m +PU m ) M hijklmn = BTi + Σ c jm.f jm +LN k + CACL(LN) hk + CICL(LN) lk +AN m + PU m + e hijklmn The model for claw disorder is the same as the first model for claw disorder M hijklm = test day production record (milk, fat, protein and somatic cells) LN k = fixed effect number l 1, 2, 3 BT i = fixed effect of Herd-Test-Day i (1,,544) f jm = regression coefficient m of fixed curve for group of calving year-calving season j (1,..,80) in number k c jm = m-th term of Ali & schaeffer function of fixed curve CICL(LN) lk = fixed effect of calving interval l in classes (1,..., 5) nested within number k AN m = random animal effect for animal m (1,, 29384) PU m = random permanent environmental effect of animal m (1,,7267) e hijklmn = random residual effect (1,, 24123) Bivariate animal model S4 Threshold model for disorders Linear model for production Test day production at time of trimming First three s Regression models with and without nesting of stage-of-effect GLIMMIX analysis PR (y hjklm =1) = θ (µ +CACL(LN) hk + BB j + LN k + DIMCL(LN) lk + b1m + b2m 2 + PU m ) PR (y hjklm =1) = θ (µ +CACL(LN) hk + BB j + LN k + DIMCL(LN) lk + b1m(ln) k + b2m²(ln) k + PU m ) Y hjklm = trait of claw disorder (1 = positive 0 = negative) PR =, θ = link function (logit and probit), µ = overall mean CACL(LN) hk = fixed effect of calving age h in classes (1,..., 4) nested within number k BB j = fixed effect of herd-visit j (1,..., 496) LN k = fixed effect of number k (k = 1,, 4) DIMCL(LN) lk = fixed effect of days in milk l (l = 1,, 8: with classes of 50 days) nested within number k PU m = random permanent environmental effect of animal m (1,,8991) M = milk yield per (23798 recorders) b1, b2 = linear and quadratic regression coefficient of claw disorder on milk yield Two separate regression models S5 focus on regressions and fixed effects cow effect added as random effect cow effect includes genetic and p.e. cow effect accounts for repetitions All s Phenotypic relationships between LAM, DD and DID and milk yield S5 - GLIMMIX frequency incidence DD milk yield in 1000 (kg) all Lac milk in 1000 (kg) 0123 incidence incidence LAM milk yield in 1000 (kg) all Lac. DID milk yield in 1000 (kg) all Lac.

7 Estimates of heritabilities for milk production trait using the RRM in the first s S2 (production part) h² Tage days in in Milch milk ' Eiweiss protein Fettfat Milch milk Zellzahl Somatic cells Genetic correlation between claw disorders and production traits for 1 st only (for the correlation between EBV, there are 52 sires 30 daughters in the first and 30 daughters in the other s) Here: High EBV for disorders more disorders! LAM DD DID WLD SU milk 0.29 * protein fat protein % * fat % scs With national EBV - S1 With our EBV -S2 r g S 3 milk protein fat protein % fat % scs milk protein fat * = significant Genetic correlations from multvariate analyses for 1 st SE are between and Genetic correlation between claw disorders and production traits for first three s production at time of trimming (for the correlation between EBV, there are 52 sires 30 daughters in the first and 30 daughters in the other s) Here: High EBV for disorders more disorders! LAM DD DID WLD SU milk protein fat protein % fat % scs With VIT EBV S1 r g S4 * = significant milk protein fat scs Genetic correlations from bivariate analyses in the first three lac. SE are between 0.07 and 0.13 Conclusions (1) Recording data at time of hoof trimming is very important to provide a database for genetic improvement of hoof health Incidence of claw diseases is high (sub-clinical cases included) Laminitis is of utmost importance Claw diseases partly have a genetic background; this can nicely be demonstrated when comparing sire families the estimated heritabilities for claw disorders in this study show that a substantial genetic variation exists; genetic improvements are possible!

8 Conclusions (2) genetic correlations with milk yield were marginal Institute of Agricultural and Nutritional Sciences Martin-Luther-University Halle-Wittenberg Group Animal Breeding D Halle, Germany H. Alkhoder, R. Pijl, and H. H. Swalve genetic improvement of hoof health will not have negative effects on milk yield moderate correlations between conformation scores and claw disorders may help somewhat but if we wish to fight claw diseases genetically, we have to do recording of the diseases (see SWE + NL) genetic improvement of hoof health is absolutely essential Thank you for your attention!