Statistical methods I 406 STATISTICAL GENETICS TO IMPROVE ROBUSTNESS OF DAIRY COWS R.F. Veerkamp, H.A. Mulder, M.P.L. Calus, J.J. Windig, and J. ten Napel Animal Breeding and Genomics Centre, Wageningen UR, PO Box 65, 8200AB Lelystad, The Netherlands SUMMARY Rapid changes in farm scale and pressures on costs and farm management together with the negative correlation between production and fitness traits have stimulated the demand for robust cows. We have defined a robust cow as: “a cow that is able to maintain homeostasis in the commonly accepted and sustainable dairy herds of the near future” which clearly contains an element of environmental sensitivity and genotype by environment interaction (GxE). Classical breeding solutions to breed for more robustness are i) avoiding inbreeding, ii) multi-trait selection and iii) allowing more natural selection. Statistical models allow more direct selection for robustness by estimating genetic correlations between environments (discrete macro scale), reaction norms describing genotypes as a function of a continuous environmental parameter (continuous macro scale) and genetic variation in residual variance which is environmental sensitivity for a large number of unidentifiable environmental aspects with a relatively small effect each (unknown micro scale). The application of these complex models is still under development, but first results indicate more GxE for fitness traits than yield traits, albeit evidence for strong re- ranking of animals is still limited, so it is mainly the magnitude of the variance which is affected by environment. These statistical models will contribute, together with classical tools and new phenotypic and genomic measurement tools, to the breeding of cows that also fit in future dairy systems. INTRODUCTION Selection for more robust cows has been an important research topic in our group. The aim of this paper is to highlight some of the work we and others have done on robustness, especially in relation to the numerical methods used to select for more robust cows. But first we discuss conceptually why we think that robustness has become important and the definition of robustness Napel et al. (2009). DEMAND FOR ROBUSTNESS Cannon (1932) first used the term homeostasis to indicate that a body continuously acts to maintain a stable internal environment by responding to external environmental stimuli. In the last two decades, there have been concerns that high-yielding dairy cows struggle to maintain homeostasis. Several studies reported unfavourable genetic correlations between milk yield and reproductive problems, locomotive problems and udder health problems (Pryce et al. 1997; Rauw et al. 1998; Royal et al. 2000), and there is general consensus that selection for milk fat and milk protein yield alone may give an unfavourable correlated response in these traits. The magnitude of these correlated responses is rather small, compared with direct effects of environmental disturbance, albeit when the effects of breeding are accumulated across years these might be substantial. The gradual reduction in genetic levels for fertility and health will put more pressure on management to maintain performance at acceptable levels. Such cows require more management attention. At the same time, the level of management is increasingly under pressure from other directions. For example, due to economic pressure, herd size is increasing and therefore the amount of labour available per animal is decreasing. The shortage of labour is aggravated by the fact that it is increasingly difficult to find suitably skilled labour. Also, pressure on