Genetic analysis of the cumulative pseudo-survival rate during lactation of Holstein cattle in Japan by using random regression models

Abstract

Longevity is a crucial economic trait in the dairy farming industry. In this study, our objective was to develop a random regression model for genetic evaluation of survival. For the analysis, we used test-day records obtained for the first 5 lactations of 380,252 cows from 1,296 herds in Japan between 2001 and 2010; this data set was randomly divided into 7 subsets. The cumulative pseudo-survival rate (PSR) was determined according to whether a cow was alive (1) or absent (0) in her herd on the test day within each lactation group. Each lactation number was treated as an independent trait in a random regression multiple-trait model (MTM) or as a repeated measure in a random regression single-trait repeatability model (STRM). A proportional hazard model (PHM) was also developed as a piecewise-hazards model. The average (± standard deviation) heritability estimates of the PSR at 365 d in milk (DIM) among the 7 data sets in the first (LG1), second (LG2), and third to fifth lactations (LG3) of the MTM were 0.042±0.007, 0.070±0.012, and 0.084±0.007, respectively. The heritability estimate of the STRM was 0.038±0.004. The genetic correlations of PSR between distinct DIM within or between lactation groups were high when the interval between DIM was short. These results indicated that whereas the genetic factors contributing to the PSR between closely associated DIM would be similar even for different lactation numbers, the genetic factors contributing to PSR would differ between distinct lactation periods. The average (± standard deviation) effective heritability estimate based on the relative risk of the PHM among the 7 data sets was 0.068±0.009. The estimated breeding values (EBV) in LG1, LG2, LG3, the STRM, and the PHM were unbiased estimates of the genetic trend. The absolute values of the Spearman's rank correlation coefficients between the EBV of the relative risk of the PHM and the EBV of PSR at 365 DIM for LG1, LG2, LG3, and the STRM were 0.75, 0.87, 0.91, and 0.93, respectively. These results indicated that the EBV of PSR could predict the genetic contribution to survival. The EBV based on the PSR of the STRM was highly correlated with that of the MTM (0.83-0.96). Furthermore, the calculation load of the STRM was lighter than that of the MTM because the rank of the matrix of the STRM was smaller than that of the MTM. These results indicated that the STRM is an appropriate model for estimating survivability by using random regression models.

Estimation of the economic value of herd‐life length based on simulated changes in survival rate

“…Therefore, the economic value of survival rate was larger in Hokkaido than in the other regions. The longevity traits for genetic estimation have been shown to be a relative risk, according to survival analysis (Sasaki et al, 2012;Terawaki & Ducrocq, 2009), and a pseudo-survival rate, according to random regression model analysis (Sasaki, Aihara, Nishiura, & Takeda, 2017;Sasaki, Aihara, Nishiura, Takeda, & Satoh, 2015). In these cases, the economic weight of longevity corresponded to the economic value per percentage survival rate.…”

Genetic correlations between the cumulative pseudo-survival rate, milk yield, and somatic cell score during lactation in Holstein cattle in Japan using a random regression model

“…Test-day SCS was calculated as follows: SCS = log 2 [(SCC)/100] + 3 (Ali and Shook, 1980). The PSR value was esti-mated according to the methods of Sasaki et al (2015). The day of cow removal from her herd was considered the last test day of the cow when the interval between the last test day of the cow and the last test day of her herd was longer than 120 d. When the interval between the last test day of the cow and the last test day of her herd was less than 120 d, the last test day of the cow was considered the day of censoring.…”

“…The dimensions of a, p1, and p2 for milk yield and SCS were the same as those in the model for routine evaluation work in Japan. The dimensions of a and p2 of PSR were the same as those used in the model in Sasaki et al (2015). e1 w is a vector of the residual effect of lactation stage group w. e2 is a vector of the residual effect of the whole lactation period.…”

“…The ETA of the RR of PHM was estimated using The Survival Kit version 6.1 (Ducrocq et al, 2010). This analysis was performed using the analytical model and parameters reported in Sasaki et al (2015). The model for PHM was as follows:…”

“…Accordingly, the genetic correlation between longevity and milk production may differ among lactation stages; specifically, genetic correlations may be higher during early stages than later lactation stages. Using a random regression model, Sasaki et al (2015) reported that the cumulative pseudo-survival rate (PSR) could reflect changes in genetic parameters during lactation. Therefore, it is possible to assess changes in genetic correlations during lactation between PSR and milk production traits and SCS using random regression models.…”