Gernand E, König S
Thuringian State Institute of Agriculture, 07743 Jena, Germany.
Department of Animal Breeding, University of Kassel, 37213 Witzenhausen, Germany.
J Dairy Sci. 2014;97(6):3953-63. doi: 10.3168/jds.2013-7830. Epub 2014 Apr 14.
The objective was to study genetic (co)variance components for binary clinical mastitis (CM), test-day protein yield, and udder health indicator traits [test-day somatic cell score (SCS) and type traits of the udder composite] in the course of lactation with random regression models (RRM). The study used a data set from selected 15 large-scale contract herds including 26,651 Holstein cows. Test-day production and CM data were recorded from 2007 to 2012 and comprised parities 1 to 3. A longitudinal CM data structure was generated by assigning CM records to adjacent official test dates. Bivariate threshold-linear RRM were applied to estimate genetic (co)variance components between longitudinal binary CM (0 = healthy; 1 = diseased) and longitudinal Gaussian distributed protein yield and SCS test-day data. Heritabilities for liability to CM (heritability ~0.15 from 0 to 305 d after calving) were slightly higher than for SCS for corresponding days in milk (DIM) in the course of lactation. Daily genetic correlations between CM and SCS were moderate to high (genetic correlation ~0.70), but substantially decreased at the very end of lactation. Genetic correlations between CM at different test days were close to 1 for adjacent test days, but were close to zero for test days far apart. Daily genetic correlations between CM and protein yield were low to moderate. For identical DIM (e.g., DIM 20, 160, and 300), genetic correlations were -0.03, 0.11, and 0.18, respectively, and disproved pronounced genetic antagonisms between udder health and productivity. Correlations between estimated breeding values (EBV) for CM from the RRM and official EBV for linear type traits of the udder composite, including EBV from 74 influential sires (sires with >60 daughters), were -0.31 for front teat placement, -0.01 for rear teat placement, -0.31 for fore udder attachment, -0.32 for udder depth, and -0.08 for teat length. Estimated breeding values for CM from the RRM were compared with EBV from a multiple-trait model and with EBV from a repeatability model. For test days covering an identical time span and on a lactation level, correlations between EBV from RRM, multiple-trait model, and repeatability model were close to 1. Most relevant results suggest the routine application of threshold RRM to binary CM to (1) allow selection of genetically superior sires for distinct stages of lactation and (2) achieve higher selection response in CM compared with selection strategies based on indicator type traits or based on the indicator-trait SCS.
目的是使用随机回归模型（RRM）研究泌乳期二元临床型乳房炎（CM）、测定日蛋白质产量以及乳房健康指标性状[测定日体细胞评分（SCS）和乳房综合体型性状]的遗传（协）方差组分。该研究使用了来自15个选定的大规模合同牛群的数据集，包括26,651头荷斯坦奶牛。记录了2007年至2012年的测定日生产数据和CM数据，涵盖第1至第3胎次。通过将CM记录分配到相邻的官方测定日期，生成了纵向CM数据结构。应用二元阈值线性RRM来估计纵向二元CM（0 = 健康；1 = 患病）与纵向高斯分布的蛋白质产量和SCS测定日数据之间的遗传（协）方差组分。在泌乳期，产犊后0至305天CM易感性的遗传力（遗传力约为0.15）略高于相应泌乳天数的SCS遗传力。CM与SCS之间的每日遗传相关性为中度至高（遗传相关性约为0.70），但在泌乳期末大幅下降。不同测定日的CM之间的遗传相关性在相邻测定日接近1，但在相隔较远的测定日接近零。CM与蛋白质产量之间的每日遗传相关性为低至中度。对于相同的泌乳天数（例如，泌乳天数20、160和300），遗传相关性分别为-0.03、0.11和0.18，这否定了乳房健康与生产力之间存在明显的遗传拮抗作用。RRM得出的CM估计育种值与乳房综合线性体型性状的官方估计育种值之间的相关性，包括来自74头有影响力公牛（女儿数>60头的公牛）的估计育种值，前乳头位置为-0.31，后乳头位置为-0.01，前乳房附着为-0.31，乳房深度为-0.32，乳头长度为-0.08。将RRM得出的CM估计育种值与多性状模型得出的估计育种值以及重复性模型得出的估计育种值进行了比较。对于涵盖相同时间跨度的测定日以及泌乳水平，RRM、多性状模型和重复性模型得出估计育种值之间的相关性接近1。最相关的结果表明，将阈值RRM常规应用于二元CM，以便（1）能够为泌乳期的不同阶段选择遗传上更优的公牛，以及（2）与基于指标体型性状或基于指标性状SCS的选择策略相比，在CM方面实现更高的选择反应。
