The estimation of genetic parameters of longevity traits in black and white cattle breed

Sprovedeno istraživanje imalo je za cilj da se procene genetski parametri osobina dugovečnosti, kao i da se odabere adekvatna metodologija i razvije optimalni model za genetsko vrednovanje osobina dugovečnosti u populaciji crno belih goveda u Srbiji. Pored genetskih parametara osobina dugovečnosti, analizirani su i razlozi izlučenja krava iz proizvodnje. Istraživanje je sprovedeno na setu podataka koji je obuhvatao podatke o dugovečnosti, razlozima izlučenja i poreklu krava, koje su se prvi put telile u periodu od 1989. do 2012. godine na Poljoprivrednoj korporaciji Beograd AD. Razlozi izlučenja su svrstani u 10 kategorija, prema srodnosti, i analiziran je njihov udeo prema nivou proizvodnje, laktaciji u kojoj su grla izlučena i udelu gena holštajn frizijske rase. Dugovečnost krava je posmatrana kroz tri pokazatelja: trajanje produktivnog života (LPL), životnu proizvodnju mleka (LMY) i broj ostvarenih laktacija (NL). Set podataka na kojem je izvršena analiza razloga izlučenja sadržao je samo grla koja su bila izlučena i imala su evidentiran razlog izlučenja, ukupno 12309 krava. Genetsko vrednovanje osobina dugovečnosti je izvršeno je upotrebom Weibull-ovog modela proporcionalnih rizika, u okviru analize preživljavanja, i linearnih modela. Weibull-ov model proporcionalnih rizika je uvažio fiksne vremenski nezavisne efekte farme, udela gena holštajn frizijske rase i uzrasta pri prvom teljenju, dok su efekti godine i sezone teljenja, kao i relativne proizvodnje mleka i laktacije tretirani kao fiksni vremenski zavisni uticaji. Otac je posmatran kao fiksni vremenski nezavisan uticaj (model oca). Primena Weibull-ovog modela proporcionalnih rizika omogućila je uključivanje u analizu i nepotpunih (cenzurisanih) zapisa. Ukupan broj krava koji se nalazio u setu podataka korišćenom u analizi preživljavanja bio je 22109, od čega su 26% činili nepotpuni zapisi. Set podataka koji je korišćen za genetsko vrednovanje osobina dugovečnosti upotrebom linearnih modela, sadržao je podatke o 16539 krava koje su imale poznato trajanje produktivnog života. Model koji je korišćen u genetskom vrednovanju osobina dugovečnosti upotrebom linearnih modela uvažio je farmu, godinu i sezonu prvog teljenja, udeo gena holštajn frizijske rase, uzrast pri prvom teljenju i relativnu proizvodnju mleka u prvoj laktaciji kao fiksne faktore, dok je sama životinja tretirana kao slučajan faktor (model individue)...The research was conducted with the aim to estimating genetic parameters of longevity traits as well as to choose adequate methodology and develop optimal model for genetic evaluation of longevity traits in the population of Black and White cattle in Serbia. Besides genetic parameters for longevity traits the culling reasons were also analysed. The research was conducted on the basis of set of data which included records on longevity, culling risks and origin of cows that started first calving in the period from 1989 to 2012 at the Agricultural Corporation Belgrade AD. The culling reason, according to their relationship, were divided into 10 categories and their share in the level of production, lactation in which animals were culled and share of Holstein Friesian breed genes was analysed. Longevity of cows was observed through three indicators: length of productive life (LPL), life milk yield (LMY) and number of realised lactations (NL). Set of data which included 12309 cows in total and on the basis of which the analysis of culling risks was conducted included only the culled animals with known reason for culling. Genetic evaluation of longevity was performed by means of Weibull proportional risk model within the survival analysis and linear models. Weibull model of proportional risk took into account fixed time-independent effects of farm, share of genes of Holstein Friesian breed and age at first calving while the effects of year and season of calving as well as relative milk yield and lactation were treated as fixed time-dependent effects. Sire was regarded as random time-independent effect (sire model). The use of Weibull proportional risk model enabled including the incomplete (censured) records into the analysis as well. Total number of cows involved in the set of data used in the survival analysis was 22109, out of which 26% were incomplete records. Set of data used for genetic evaluation of longevity traits by applying linear models contained data on 16539 cows whose length of productive life was known. Model used in genetic evaluation of longevity traits by applying linear models involved the farm, year and season of first calving, share of genes of Holstein Friesian breed, age at first calving and relative milk yield in first lactation as fixed factors, while the animal itself was treated as a random factor (animal model). The average length of productive life for all animals included in the analysis was 1267 days (41.7 months), that is 1300 days (42.7 months) for the animals with known length of productive life and 1170 days (38.4 months) for censored records..