Rates of genetic improvement in dairy cattle breeding programmes have increased substantially in the last decade, not only in Europe where there has been substantial immigration of North American stock. More accurate statistical evaluation procedures, including the use of best linear unbiased prediction (BLUP) with the animal model, have been introduced; higher selection intensities have been achieved, particularly on production traits, through better organized and focused schemes; and developments in multiple ovulation and embryo transfer (MOET) have been both a stimulus and to some extent a cause. Continued rapid improvement can be expected as research and development enables more accurate and timely use to be made of recording data, because there is evidence that heritability values for milk production are rising, perhaps because of better cow management, and as competition among breeders internationally increases. There are a number of consequent challenges to geneticists and breeders. Attention will have to be given to maintaining the fitness of very high producing animals by recording health, fertility, longevity and parlour traits, and using the results of research to give them appropriate weight in selection decisions. Developments in molecular methods and in reproductive technologies present new opportunities, but are unlikely to do more for some time than complement progress from conventional selection on the important production traits.

The objective of any well designed progeny test programme is to identify those individuals that have inherited the favourable genes from his parents. Bulls that enter a progeny test programme have been highly selected on a set of selection criteria. The criteria vary among organizations based upon their breeding philosophy and their prediction of the future economic value of various traits. The accuracy of choosing among this highly selected group is quite low. Increasing the accuracy of selection before progeny testing is the greatest potential application of genetic marker technology. Markers associated with traits of importance can greatly enhance traditional selection methods by increasing the prospect of an individual having the desired characteristics. Genetic marker-assisted selection can greatly increase the actual genetic merit of traits of economic importance

In principle, food intake by lactating dairy cows might be limited by rate of eating, capacity to produce saliva, rumen or intestinal capacity, rates of digestion and passage of food residues, or by factors such as chemicals and osmolality in the rumen. It is likely that a combination of these is used by the central nervous system to control intake. In practice, the positive relationship between digestibility and intake of forages means that high-quality material must be offered to cows during lactation if a reduced level of concentrate supplementation is envisaged but undue mobilization of body reserves avoided. Better supplementation is possible, both in terms of more frequent, smaller allocations, and of more appropriate concentrate composition with regard to the composition of the basal forage. Intake of maize silage dry matter is usually greater than for grass silage and a mixture of the two, perhaps offered in free or restricted choice, is likely to become more widely practised. Changes in body reserves are difficult to monitor and this has made it difficult to resolve the question of the magnitude of diet/environment interactions, i.e. is the same breeding and selection programme appropriate for those animals which are intended to be pushed to high yields in intensive systems and those intended to be given principally forages in semi-extensive systems?

Only as recently as 1992 was the digestible crude protein approach to the calculation of protein requirements for ruminants abandoned in favour of a scheme in which metabolizable protein (MP) became the nutritional currency (Agricultural and Food Research Council, 1992). Application of this scheme requires reliable methods to characterize foods in a number of ways and whilst there are deficiencies in current methodology for this purpose there is now a sufficient body of information available to sustain use of the scheme. Evaluation of the MP approach shows it to be a considerable advancement on what went before but still with some inadequacies. More attention needs to be given to the impact of food constituents on digestive processes within the rumen. Tests of the relationship between predicted MP supply and dairy cow performance show that rules are needed to predict partition of available MP between alternative routes of use. With such rules it will become more likely that a useful response prediction system can be developed. Ideas are put forward as to how some progress might be made to develop frameworks which will bring together ideas about responses to protein and energy yielding nutrients. Clarification of partition rules will be important in promoting improved feeding strategies for cows of different genetic index for milk production. However, it is felt that, given adequate partition rules, the MP approach applies equally to high genetic index and low index cows as regards calculation of requirements. A concern is expressed, however, that the current UK research environment may not be conducive to the open sharing of information which will be important to make rapid progress towards the development of nutrient response frameworks. A technology transfer system is needed which will enable advances in knowledge to be organized for translation into practice.

Ongoing improvements in the genetic merit of United Kingdom (UK) dairy cows as well as changes in the national milk market will increase awareness of and provide focus on the importance of balanced nutrition, particularly with respect to energy and protein. Within the UK, energy is rationed according to the metabolizable energy (ME) scheme, proposed in the 1960s and implemented in the late 1970s. Whilst a significant improvement over previous rationing systems, deficiencies in the scheme have been identified. These relate to an inability to predict animal response including the importance of nutritionally induced changes in milk composition, and to represent adequately body tissue mobilization in early lactation and the control of nutrient partitioning during the phase of tissue repletion. Some of these issues are addressed in this paper. These include the problems likely to occur in formulating rations for dairy cows producing in excess of 50 kg milk per day during significant parts of the lactation, which leads to an appraisal of some of the concepts embodied in the current ME system. By reference to appropriate experimentation, the importance of recognizing the nutrient composition of ME as well as the significance of nutrient metabolism in the portal drained viscera and the liver is established. The rôle of added fat within the diet is discussed, whilst the quantitative importance of mobilized body tissue to whole body energy metabolism and the interaction between specific nutrients is recognized. Finally, it is concluded that a complete revision of the ME system is unlikely at this stage and therefore it will be prudent to deploy available resource to modify the existing ME model in line with some of the inconsistencies and weaknesses that have been identified.

A comparison of UK (Agricultural Research Council; ARC) and USA (National Research Council; NRC) feeding standards for dairy cows is made. The two standards appear similar in terms of food energy required for milk production, even though the ARC uses metabolizable energy and the NRC uses net energy. The two systems have the same basic approach to calculating protein supply, although ARC incorporates information on level of food intake in calculating rumen microbial protein synthesis and in estimating undegraded food protein. The maintenance requirement for protein is much less with the ARC system because NRC assigns a large maintenance charge to metabolic faecal nitrogen. The result is that ARC has lower dietary protein recommendations than NRC. Milk production response to incremental additions of energy (grain) or protein diminishes as the point of maximum response is approached. The most profitable level of energy and/or protein supplementation is often below the supplementation level required for maximum milk production. More emphasis must be given to lactation studies for information to improve accuracy of the feeding standards and for construction of milk response curves to incremental additions of supplements.

In recent genotype X environment studies between New Zealand and Canada, yields of fat and protein from Holstein Friesian (HF) cows were similar for daughters of sires from both countries, but Canadian daughters produced larger volumes of milk (6500 kg v. 3300 kg) and were taller. There was no interaction between the sire's national strain and the daughter's environment but there was a significant interaction between-individual sires within-strain and the daughter's environment so that the correlations between the sire proofs in the two countries were approximately half of the expected values.

In New Zealand, milk production by cows of high or low genetic merit has been found to be affected similarly by differences in stocking rate, level of feeding and body condition at calving, providing no evidence of a large genotype X management interaction, although there is some evidence of an interaction between the effects of breed size and feeding, with smaller breeds being affected less severely by lower feeding level and clear evidence of genotype by milking management interactions. Over the past 40 years the need for pre-milking stimulation in New Zealand cows has been reduced or eliminated by inadvertent selection. Other recent evidence suggests that progress for genetic tolerance to once-daily milking may be possible.

It can be concluded that interactions between group-average genotype and nutritional environment are not likely to have important effects on milk solid yields, at least within the HF breed, within the range of conditions normally found in temperate dairy systems, and provided that the environment is imposed from birth. However, there are likely to be significant interactions between the genotype of individual sires and environment and thus appropriate assessment and ranking of sires in each environment is essential.

Genotype by environment interaction was investigated for a range of traits. Selection (S) and control line (C) cows, housed and managed at the Langhill Dairy Cattle Research Centre, were offered ad libitum complete mixed diets, with proportions (in total dry matter) of concentrates, silage, brewers’ grains of either 20:5:75 (LC; 1.0 t concentrate per annum) or 45 :5 :50 (HC; 2.5 t concentrate), over a full lactation. No diet X genetic line interactions were observed for a range of traits (production, food intake, live weight and condition score) when compared as means. However, regression coefficients of protein yield (P < 0.05), milk yield (P < 0.10) and protein percentage (P < 0.10) on pedigree index for fat plus protein yield were different between LC and HC. Genetic correlations between HC and LC were high for most traits, and only for fat yield was the value below 0-80. Because the rate of increase in energy output from selection is unlikely to be matched by the rate of increase in intake during the first part of the lactation, it was expected that intake capacity is likely to become more important in the future, regardless of the feeding system. For this reason, it is suggested that selecting for a lower live weight whilst simultaneously selecting for increased yield (an approach which has been advocated in some countries) should be treated cautiously. Especially on low input diets increases in yield as a result of continued selection (on higher input diets) may be limited by the rate, or extent of tissue mobilization; under these circumstances health and fertility problems might be the first indicators of such limits being approached.

Major increases in the rate of genetic improvement in the dairy herd have been obtained in the United Kingdom and the Republic of Ireland since the mid 1980s. The implications of increases in genetic merit and the possible consequences of genotype X nutrition interactions on the efficiency of milk production systems are reviewed. The majority of previous studies with dairy cattle of moderate genetic merit suggest little evidence of genotype X nutrition interactions across a range of nutritional and management regimes, with higher milk production of high merit cows largely accounted for by effects on nutrient partitioning. However, more recent results suggest a significant re-ranking of sires when evaluated under either intensive feeding systems or in systems with a high reliance on grazed pasture.

Under intensive feeding systems higher animal performance has been obtained with high merit cows across a range of concentrate inputs and feeding systems. However, recent results from the Langhill studies provide the first tentative evidence of a genotype X nutrition interaction, with significantly different regression coefficients between genetic merit (as assessed by pedigree index) and milk production, under either low or high forage diets. The implications of these results are that high merit cows may be unable to express their full genetic potential for milk production when offered a high forage (or low energy density) diet. Consequently, the influence of other factors which have a major effect on voluntary food intake, and hence nutrient intake, e.g. forage dry matter content and forage digestibility, may be relatively more important with high genetic merit dairy cows.

It is concluded that higher milk production in high merit cows is largely attributable to variation in partitioning of nutrients, rather than to changes in food intake or digestive efficiency. Consequently, it is important fully to assess the animal performance and welfare implications of maintaining high genetic merit dairy cows under systems which may limit nutrient intake, e.g. under high forage regimes involving a high reliance on grazed or conserved forage. The major challenge for research and dairy herd management is to increase food intake with grass- and/or grass silage-based diets, thereby reducing the need for high levels of concentrates input to prevent excessive body condition loss with high genetic merit dairy cows in early lactation.

Milk yield is ultimately constrained by the number of mammary secretory cells and the amount produced by each cell; increasing yield during early lactation is associated with increased output per cell, whilst decreasing yield during declining lactation correlates with reduced number of cells. Overall, yield is highly correlated with mass of secretory tissue but strategic control of lactation involves a galactopoietic complex of hormones including GH, prolactin and oxytocin acting to ensure efficient milk ejection and maintenance of secretion. Fine control, on the other hand, is achieved locally within the mammary gland by an autocrine mechanism which matches supply of milk to demand. The feedback inhibitor of lactation (FIL) is a milk protein which is inhibitory to secretion, so as milk accumulates between milkings, secretion rate gradually falls. The more frequently FIL is removed by milking, the greater the overall secretion rate. Storage of milk occurs within secretory tissue (alveolar milk) but also in the cistern (cisternal milk); FIL is effective in alveolar milk but not in cisternal milk, because it is then remote from its site of action. Therefore predictions would be that for a given mass of secretory tissue, large-cisterned cows should produce more milk, be more tolerant of infrequent milking but be less responsive to frequent milking. Methods developed by us for determining cisternal and alveolar milk storage spaces have obtained statistically proven support for the latter two predictions in dairy cows; the first prediction has been proven in goats. It has been shown that cisternal milk fraction increases during the course of lactation and with increasing parity. It is now intended to investigate the feasibility of incorporating storage characteristics into future selection strategies.

This paper considers the welfare implications of breeding, feeding and managing dairy cows for increased productivity. The physiological demands of lactation for the high genetic merit dairy cow are exceptional, less in their intensity and more in their duration. Most welfare problems can be attributed less to productivity per se than to systems of feeding, milking and management that are unsuitable to the genotype of the high yielding cow. These include the inadequacies of wet grass silage as a staple food and the abnormal restriction of cows to twice-daily milking. It is also necessary to consider welfare problems that may arise from the application of new technologies to increase milk yield or to increase the rate of genetic progress. These may be related to the process itself or to its consequences. It is proposed that the commercial exploitation of these new technologies should be controlled by legislation similar to that for the protection of animals used for scientific purposes which applies a cost-benefit analysis to determine whether the cost to the animal can be justified in terms of any potential benefit to society.