Analysis of survival in dairy cows with supplementary data on type scores and housing systems from a region of northwest Germany

In survival analysis, type traits can be included as covariates to evaluate their use as predictors for survival. One problem in such an analysis is t he availability of suitable data. Whereas data on the length of productive life (LPL) of individual cows can be retrieved from milk recording data, fo r type traits, all cows in the population must be scored for type at least once. In the present analysis, a dataset from the Osnabruck region in north western Germany, which fulfilled this requirement in recent years, was used . Data consisted of 169,733 cows with information on LPL for calving years 1980 to 1996 (dataset I) and of 39,233 cows with information on LPL and typ e for calving years 1990 to 1996 (dataset II). A further dataset (III) cont ained 43,116 cows from calving years 1987 to 1996 and included information on the housing system for each herd. The basic model included stage of lact ation, relative production within herd, change of herd size, and year-seaso n as time dependent effects; age at calving as a time-independent effect; a nd herd-year-season and sire as random effects. Other effects (information on type, housing system) were included additionally. For dataset II, the sc ores for 15 linear type traits were also included as corrected phenotypic v alues, estimated breeding values, and residuals from a previous BLUP analys is. The package Survival Kit 3.0 was used for all analyses. The results ind icate a moderate heritability of 0.17 and 0.18 for true and functional LPL (dataset I). Almost all type traits analyzed (dataset II) exceeded a 0.001 level of significance in their effect on survival. The strongest relationsh ips between survival and type were found for udder depth, fore udder attach ment, and front teat placement. The main result from the comparison of hous ing systems (dataset III) was that bedding has a positive effect on surviva l.