Improving the efficiency of nutrient utilisation and the predictability of product composition in the lactating dairy cow : the Underwood lecture for 1996

Abstract

Proc. Aust. Soc. Anim. Prod. 1996 Vol. 21 THE UNDERWOOD LECTURE FOR 1996 IMPROVING THE EFFICIENCY OF NUTRIENT UTILISATION AND THE PREDICTABILITY OF PRODUCT COMPOSITION IN THE LACTATING DAIRY COW D.E. BEEVER Centre for Dairy Research (CEDAR), Dept of Agriculture, The University of Reading, Earley Gate, Reading RG6 6AT The UK dairy industry as part of the larger European market has experienced considerable change over the last decade or so and this situation is expected to continue. Following the introduction of quotas on milk production in 1984, a payment structure based on milk constituents was imposed with a price ratio of protein to fat of approximately 0.9:1. This differential in price now exceeds 3:2, reflecting oversupply of fat and an increased market demand for protein by milk processors. Consequently, all farmers are limited by a fat production quota, and the penalty on overproduction of milk, after correction to the fat base level, can substantially exceed the prevailing milk price (current penalty =1.2x milk price). Thus there are no incentives to overproduce milk, and farmers respond, either by the annual leasing or purchase of quota (current prices; approximately 50% and 300% of milk price), restricting output per cow by nutritional means or reducing the number of cows milked. In 1994, the Milk Marketing Board was disbanded, and replaced by a free market where individual dairies negotiate with farmers, offering incentives according to the quantity and composition of milk they require. Currently the market for milk is buoyant, although most predictions indicate this position will not be sustained and farmers anticipate lower returns in the longer term. Feed costs still constitute a major part of the costs of milk production whilst feed composition can have a significant effect on milk composition. Thus there is renewed-interest in opportunities to improve the overall efficiency of nutrient utilisation by the cow, whilst increasing milk protein content and maintaining or possibly reducing milk fat content and yield. The introduction of feeder wagons on many farms has reduced the reliance on purchased concentrates, and increased on-farm use of commodity feeds (e.g. soya, wheat, maize gluten) with increased awareness of the importance of forage quality, principally grass and maize silage. If such changes were not sufficient, the UK industry is now at the centre of the Bovine Spongiform Encephalopathy (BSE) issue, and it is premature to predict the implications of such on the UK dairy and beef industries. This paper will consider the implications of some of these changes and attempt to relate current research effort to the practical feeding of dairy cows for quality milk production. Nutrient intake Given that a significant proportion of UK milk is produced from conserved forage, the importance of optimising silage intake is obvious. Grass silage remains the principal source of conserved forage, with most of that fed to dairy cows being produced from heavily fertilised grass swards cut as primary growths in early May. Most of this silage will be clamped, with increased attention being paid to ensiling conditions as demonstrated by Aston et al. (1994) who, in comparing four silages fed to dairy cows, concluded that the ensiling process was considered to have a greater effect on silage intake than a reduction in grass quality, as influenced by factors such as inclement weather at the time of harvesting. However, despite a comprehensive characterisation of the silages, Aston et al. (1994) were unable to identify those factors responsible for the large differences in dry matter (DM) intake noted when the silages were fed without supplement to lactating dairy cows. Thus, prediction of silage intake remains a research goal, and it may now be opportune to abandon the empirical approach to feed intake regulation and establish the importance of the mechanisms involved. At the same time, feeding maize silage to dairy cows has increased dramatically. In 1990,30,00Oha of maize were grown for ensiling; by 1995 this had increased to over 110,OOOha and with the introduction of early maturing varieties plus improved knowledge on the agronomy of maize, the production of maize silage has spread to more northerly latitudes of Britain. In studies at CEDAR, the effect of replacing 33% of the grass silage component of the diet with maize silage on increased forage dry matter (DM) intake and improved milk output was established, and more recently Phipps et al. (1995) found that this response was maintained as the proportion of maize silage in the forage component of the ration was 1 INTRODUCTION Proc. Aust. Soc. Anim. Prod. 1996 Vol. 21 increased to 75% (Table 1). Consequently, many farmers include maize silage in excess of 60% of the forage component of the ration. Also, Phipps et al. (1995) examined other potential replacements for grass silage, and established that ` forage' DM intake was stimulated in every case as grass silage was partially replaced within the diet (Table 1). In most instances, these changes were accompanied by improvements in milk yield and alterations in milk composition, the reduction in milk fat content associated with brewers grains inclusion being particularly interesting. A similar effect was observed by Smoler et al. (1995) when barley dark grains were included in the diet and the mechanisms of this effect are currently being investigated. Table 1. The effect of replacing part of the grass silage component of the ration of lactating dairy cows with ` alternative forage' sources on forage intake, milk yield and milk composition. FW, UW, BG, FB, MS refer to fermented whole crop wheat silage, urea treated whole crop wheat silage, brewers grains, fodder beet and maize silage respectively, all at 33Og/kg total forage DM, and MSH refers to maize silage inclusion at 75Og/kg total forage DM, with grass silage (GS) comprising the remainder of the forage component in each diet (i.e. 670 or 25OgIkg total forage DM respectively) Despite such opportunities with forage alternatives, grass silage will remain the principal forage fed to housed dairy cows. However, in response to impaired feed intakes often recorded on farm, and a realisation that the nutritional value of grass silage appeared to be higher in Holland where grass is extensively wilted prior to ensiling, UK farmers are abandoning the direct ensiling of low DM grass. The wilting of cut grass prior to ensiling is increasing and many farmers are investing in wide rakes to spread the grass in order to increase the rate of moisture removal. Undoubtedly, such practices are leading to increased silage intakes, but much remains to be understood about the ensiling process if the ultimate predictability of silage quality is to be achieved. There are over 100 silage additives on the UK market, but farmers have difficulty in establishing which of the manufacturers claims are likely to be of financial benefit to their businesses. However, until greater attention is paid to the chemical composition of the crop at harvesting and ensiling, it is unlikely that sustained advances in silage feeding value will be achieved. Nutrient digestion and absorption Introduction of the UK metabolisable protein (MP) scheme in 1992 was the culmination of many years research effort into the factors influencing protein utilisation by ruminants, albeit heavily biased towards events within the gastro-intestinal tract, with scant regard to post absorptive processes. Such changes have increased awareness of variation in the nutritional value of different protein sources, in terms of their relative abilities to support microbial protein synthesis or augment small intestinal protein supply. The proposals were published for practical use by Alderman and Cottrill (1993), as other countries in Europe and the USA were presenting their own recommendations. Whilst all systems are conceptually similar, they vary in detail and some of the approaches adopted. This was inevitable and the lack of a standardised procedure, at least within Europe, is regrettable. Consequently, where the UK system was found to be inadequate, other systems have been considered. Currently, the Cornell Net Carbohydrate and Protein System (CNCPS) is gaining some popularity within the UK, despite the feed database having little relevance to UK feeds. There is no proof, however, other than anecdotal evidence, that CNCPS is superior to other models in predicting animal performance. One major development within the UK system was recognition of the importance of energy:protein interactions, albeit only with regard to ruminal metabolism. Fermentable metabolisable energy (FME), defined as the energy available within the rumen to support microbial metabolism, is assumed to be dietary ME intake less the energy content of dietary lipids and fermentation end products (e.g. as in ensiled feeds), both being considered to be unavailable for microbial metabolism. In principle this 2 hoc. Aust. Sot. Anim. Prod. 1996 Vol. 21 definition is acceptable but the values derived will be influenced by the value ascribed to the ME content of the feed. In the study of Beever et al. (1996), predictions of the ME content of 2 maize silages by 8 independent laboratories showed a 2 MJ/kg DM range, and few of the estimates bore similarity with those obtained when the same diets were fed to lactating dairy cows, and corrected for level of feeding. All laboratory estimates indicated higher ME values for the later harvested maize silage when, in fact, the in vivo estimates for this diet were lower, despite a higher starch content, due to the lower gross energy content associated with a lesser extensively fermented crop. As to the deductions associated with the contents of lipids and fermentation products, in theory this should be relatively straightforward. However, such nutritional entities require extensive laboratory analysis, and thus current practice is to predict them from other feed parameters, as suggested by the equation based on the oven DM content of silage proposed by ADAS (1992). To date no in viva estimates of FME are available and not until such are available, will progress be made towards reliable predictions of FME values on a routine basis. One further point relates to the failure to discount dietary proteins which are unavailable for degradation in the rumen. On many diets, this omission will be inconsequential, but on those containing significant amounts of feeds such as fish meal and protected soya to specifically augment small intestinal amino acid supply, the effect on FME estimates may be considerable. For fishmeal, the quoted FME content is 12.1 MJ/kg DM, yet with an agreed digestible undegradable protein (DUP) content of 344g/kg DM, it is not possible for FME content to exceed 4.1 MJ/kg DM. In support of the MP system, feedstuff analyses now include estimates of the respective contents of effectively rumen degradable protein (ERDP), and DUP. Both are important in rationing protein for ruminants, but are routinely derived from relationships which were developed from studies where the in sacco digestion procedure was used to estimate the relative proportions of the different protein fractions. As indicated by Beever and Cottrill (1993), intra laboratory variation in such procedures can be considerable, even when standardised procedures are used, and current experience is that the technique is not sufficiently sensitive to detect differences which may have considerable biological importance. Once again, the in vivo data base, especially with respect to high producing dairy cows, is inadequate and must be improved before robust predictive equations can be established for routine laboratory use. Unfortunately, many laboratories ignored this concern and continue to report data that often serves to confuse rather than clarify. Unfortunately there has been little concerted effort to assess the degradation characteristics of the carbohydrate components of feedstuffs, despite recognition of the importance of readily fermentable carbohydrate (sugars and starches) in the diet, or the fact that different starch sources, especially those which have been processed, will have different rates and extents of digestion in the rumen. There is some interest by commercial companies in assessing the digestion characteristics of the starch component of maize silage, and results have indicated that they are influenced by stage of harvest and subsequent ensiling conditions. Knowledge of the in vivo ruminal digestion of starch, and the level of dietary starch likely to escape rumen degradation in dairy cows, remains limited, although this does not prevent laboratories and consultants indicating the relative proportions of ruminally-available and -resistant starch. The importance of energy and protein synchronisation in the rumen was elegantly demonstrated by Rooke et al. (1987), where despite modest levels of grass silage fed to non lactating cows, the consequences of nutrient losses during the ensiling process were quantified. The results (Table 2) show that on the control diet, rumen ammonia levels were relatively high and both yield and efficiency of microbial protein synthesis were low. Ruminal supplements of casein or urea (by infusion) had no discernible effects on microbial yield, whereas infusion of glucose significantly increased microbial protein synthesis, and this accounted for the increased flow of non ammonia nitrogen to the small intestine, whilst rumen ammonia concentrations were significantly reduced. However, the largest response occurred when glucose and casein were co-infused, with a 50% improvement in microbial protein synthesis compared with the unsupplemented silage. Such data permit quantification of the consequence of ensiling in terms of reduced nutrient availability and establish the need to restrict this loss by improved ensiling methods or to minimise the effect on the animal by the use of appropriate supplements. One other area of concern is the inability to predict the outcome of rumen fermentation in terms of total volatile fatty acid (VFA) yield and composition. Given propionate is the primary substrate for glucose in the ruminant, and that the dairy cow yielding 35 kg milk/d has a minimal glucose requirement of 2.5 kg/d (MacRae et al. 1988), the need to establish how much of this demand can be met from propionate or small intestinally derived glucose, with recourse to the use of gluconeogenic amino acids 3 Proc. Aust. Soc. Anim. Prod. 1996 Vol. 21 as necessary, is self evident. Equally, acetate and butyrate are important precursors for the synthesis of milk fat, supported in early lactation by fatty acids derived from mobilised body fat. The need to control milk fat synthesis has been discussed, whilst the ability to manipulate the relative proportions of unsaturated and saturated fatty acids in milk is likely to become more important. Consequently, the need to predict the relative quantities of lipogenic VFA arising from rumen fermentation can not be ignored. Furthermore, partition of hexose disposal within the rumen between that used directly to support microbial growth, that which is fermented to supply essential ATP to support microbial maintenance and that considered to be in excess and thuseither fermented or incorporated into microbial polysaccharide, will have a major effect on the relative yields of individual VFA (Beever 1993). Table 2. The effect of intraruminal infusions of nutrients on nitrogen metabolism in the rumen of cattle receiving grass silage. Cas. and Glu. refer to casein and glucose infusions (OMADR = organic matter apparently digested in the rumen) Despite considerable interest in the manipulation of rumen VFA proportions some 15 to 20 years ago, using ionophores (Chalupa 1979) or buffer salts (to increase rumen dilution rate, Harrison and McAllan 1979), there is little evidence of such technologies being applied in the European dairy industry. The ionophore monensin, which has been successfully used within the beef industry throughout the world, has been undergoing dairy efficacy trials within Europe and the USA. The aim of this co-ordinated study was to consider its effects on reducing milk fat content, whilst possibly increasing milk protein content, and the consequence of such changes on overall efficiency of nutrient utilisation; all being considered as possible consequences of alterations in the glucogenic:lipogenic VFA ratio. Limited results are available from this study, and the future of further research is highly dependent on a possible European directive to ban the use of all feed additives, including ionophores. Nutrient supply and metabolism Following absorption, rumen derived VFA enter the ruminal vein and are transported via the portal vein to the liver. The rumen wall is metabolically active, and whilst acetate is absorbed unchanged, both propionate and butyrate may be extensively metabolised to lactate and Is--OH butyrate respectively, during absorption. Within the portal vein, rumen VFA are associated with amino acids and glucose (if any) absorbed from the small intestine, plus small quantities of VFA derived from hind gut fermentation of potentially degradable carbohydrates which escaped ruminal digestion. Studies by Reynolds et al. (1995) and others indicate that amino acid metabolism withi n the portal drained viscera may be significant in response to the extent of protein turnover which occurs in such tissues, and will reduce amino acid supply to the liver compared with the quantity absorbed from the intestines. Amino acid catabolism may also be extensive within the liver. From nutrient balance studies across the liver, it would appear that the quantity of individual amino acids, leaving the liver in the free form, with the possible exception of the branched chain amino acids may be substantially less than the quantity entering the liver. This may be related to the synthesis and subsequent export of serum proteins or peptides from the liver, or to homeostatic mechanisms operating within the liver to regulate amino acid output to peripheral tissues. Additionally, part of this loss may be associated with the role of the liver to remove ammonia, by conversion to urea. Ammonia entering the liver will be principally of ruminal origin, under conditions where nitrogen availability exceeds microbial demands. In normal feeding conditions, hepatic ammonia load will not be excessive and unlikely to fluctuate widely with respect to feeding pattern, and it would appear that the liver is capable of quantitatively removing ammonia as urea, with ammonia-N removal and urea-N appearance being in stoichiometric balance. However, when ammonia loading increases, and especially when the pattern of ammonia supply fluctuates, as illustrated by Wilton (1989) 4 Proc. Aust. Soc. Anim. Prod. 1996 Vol. 21 in growing cattle meal fed on grass silage, urea-N output often exceeds ammonia-N removal, suggesting the involvement of an alternative source of nitrogen which donates one NH group to combine with one ammonia molecule. This phenomenon was reported by Maltby et al. (1991), and using N-l 5 labelled ammonia, Lobley et al. (1995), demonstrated that a substantial part of the urea synthesised in the liver was derived from non-ammonia (i.e. non N-15 labelled) sources. The extent to which this occurs in normal feeding practice is difficult to quantify, but it is possible that it could occur in animals receiving significant quantities of grazed or conserved grass. The implications of such on peripheral amino acid supply are obvious, and may, in part, explain the impaired utilisation of protein often recorded on grass based diets fed to lactating or growing ruminants. One further area of importance with respect to liver metabolism is the provision of glucose to peripheral tissues. Gut tissue is a net utiliser of glucose, albeit a significant part of this may be derived from arterial supply. Nonetheless, under most feeding conditions, the gut makes no net contribution of glucose to portal supply, and the animal relies almost exclusively on the hepatic conversion of ruminally derived propionate to glucose which is then partitioned to the productive tissues of the body. At all times, the animal needs to maintain circulating glucose levels within reasonable limits, and if required, gluconeogenic amino acids will be catabolised within the liver. As to the quantitative extent of such processes, it is difficult to provide definitive data, but with average yield of milk continuing to increase and some cows now producing in excess of 60 litres/d at peak, glucose supply becomes critical. Based on the estimate provided by MacRae et al. (1988), that one litre of milk requires a minimum of 70g glucose, this equates to an overall glucose requirement in excess of 4 kg/d. With an estimated maximum digestible organic matter intake of 17-18 kg/d, this suggests, after due allowance for energy lost as heat and methane during the processes of ruminal digestion, that almost 30% of the energy absorbed from the gastro-intestinal tract of a dairy cow must ultimately be supplied to peripheral tissues as glucose. Given that most of this will be derived from propionate, the need to optimise ruminal fermentation is apparent, as are the possible benefits which could be achieved either by feeding diets which promote propionate fermentations or the use of appropriate rumen manipulants. As research attempts to refine dairy cow feeding in order to increase the efficiency of conversion of dietary nutrients into milk constituents and improve the predictability of milk composition, the importance of a better knowledge of metabolism in both the portal drained viscera and the liver will be realised. Suffice to conclude, at this stage, that the techniques for such studies are only now sufficiently robust that reliable arterio-venous differences of metabolites including individual amino acids can be obtained. Estimates of oxygen consumption across the portal drained viscera and the liver indicate both tissues can each account for 25% of whole body oxygen consumption, and when combined with a further 25% associated with mammary metabolism, the overall importance of such tissues to nutrient utilisation in the lactating dairy cow is apparent. Whole body metabolism The UK metabolisable energy (ME) system will predict overall energy utilisation, albeit for long term (i.e. substantial parts of the lactation) rather than short term situations. It does not however predict milk composition and how this may be affected by the composition of metabolites derived from the degradation of ingested feedstuffs. In this regard, the study of Thomas et al. (1987) provided an excellent example where increasing digestible energy intake by increasing concentrate:forage ratio in the diet gave milk energy outputs which failed to meet those predicted by the ME system (for details, see MacRae et al. 1988). In response to extra concentrates, milk yield increased but reduced milk fat content and hence yield, attributable to the changed nature of the diet, were responsible for the impaired milk energy output. Equally, Sutton et al. (1985) demonstrated that feeding the same diet in 2 or 6 meals per day markedly influenced milk solids yield. With 2x per day feeding (Table 3), increasing concentrate:forage ratio caused major reductions in both milk fat content and yield (-45% and -34% respectively of the values observed on the higher forage diet). In contrast, when the same diets were fed 6x per day, the reductions noted on the high concentrate diet were considerably less (content, -24%, yield, -22%), an effect which the authors attributed to changes in the pattern of insulin release and resultant circulating concentrations. Feeding 2x per day induced large pulsatile increases in insulin concentrations, which were considered to partition more lipogenic precursor towards body tissue synthesis, whilst insulin levels were more attenuated in the 6x per day feeding regime, and the drive towards body tissue synthesis was less ` pronounced. Further evidence of the failure of the ME system to accurately partition dietary ME to milk energy is provided by the calorimetry study of Cammell et al. (1992), which considered the effect of 3 levels of 5 Proc. Aust. Soc. Anim. Prod. I996 Vol. 21 concentrate supplementation on energy utilisation throughout the first 29 weeks of lactation. In early lactation, all cows lost body energy (Figure l), but this was greatest on the medium concentrate fed animals. This effect was maintained as lactation progressed with. cows on low and high concentrates achieving positive energy balance before those on the medium level. These responses can be reconciled by examination of differences in the respective milk energy outputs which were greater between low and medium concentrate than between medium and high levels. Thus, although medium concentrate provided more ME than the low level, cows mobilised more tissue and thus supported increased milk energy output. The mechanisms involved can not be readily identified, but it is likely that significant changes in the nature of the end products of digestion occurred as a consequence of the increased level of concentrate inclusion. Whether the observed responses were a direct manifestation of these changes or indirectly associated with changes in the endocrine status of the cows due to changed nutrient supply, it establish from the study cited. 6 Proc. Aust. Soc. Anim. Prod. 1996 Vol. 2 I One other issue concerns the effect of body composition on nutrient utilisation by the cow. In a study involving 54 multiparous cows receiving the diets used by Cammell et al. (1992), serial slaughtering from immediately post-calving until lactation week 29 allowed changes in body composition to be quantified. From this Gibb et al. (1992) found that in post-calved cows, fat accounted for 69% of emptybody energy, with 60% attributable to the carcass, 20% as visceral and omental fat and 11% associated with kidney, perinephric and mammary tissues. Total body protein on the other hand was located principally in the carcass (65%), along with head, hide and feet (16%) and blood and intestinal tissues (5% each). This study also demonstrated that by lactation week 8, body energy mobilisation averaged 1.6 GJ for all cows, compared with a post calving assessment of body energy of 6.3 GJ, indicating 25% of initial body energy had been mobilised. Furthermore, 90% of this loss was attributable to body fat, with mobilised protein making only a small contribution. Assuming an average efficiency of utilisation of 0.84 for mobilised tissue (Alderman and Cottrill 1993), equates to an increased availability of net energy of 1.35 GJ, representing an additional 480 litres of milk of average composition. All cows in the study lost similar amounts of body fat by week 8-l 1 of lactation, but thereafter, body fat as well as protein were repleted more rapidly with the high concentrate cows. In contrast, the low concentrate cows showed only marginally increased levels of body fat and energy by week 29. Examination of lipogenesis and lipolysis in adipose tissue taken at the time of slaughter confirmed these effects (Walsh et al. 1992). On medium and high levels of concentrates, there was a virtual cessation of lipogenesis in both subcutaneous and omental tissue until lactation week 11, whilst initiation of this process was delayed 3 more weeks on the low concentrate diet. At the same time, tissues from all cows showed accelerated lipolytic rates, and these were maintained even when lipogenesis had recommenced. Compared with major changes in fat metabolism, it is interesting to speculate on the possible metabolic fate of body protein in the postpartum cow. The data of Gibb et al. (1992) as well as that of Sutter et al. (1994), which specifically examined energy and protein. metabolism in the first 8 weeks of lactation of cows receiving adequate or suboptimal levels of dietary protein, suggest that net protein catabolism is neither extensive nor extended in such animals. Sutter et al. (1994) found all cows approached net nitrogen balance by lactation week 3, and thereafter small net nitrogen balances were observed, whilst Gibb et al. (1992) reported significant increases in the weight of specific tissues such as liver, stomach and intestines, all of which have relatively high protein contents. Thus it is suggested that during early lactation, redistribution of protein within the body is of greater significance than protein catabolism to supply energy or glucose to support milk synthesis. Consequently, it may be appropriate to consider the immediately post-partum cow as having both lactational and growth demands, and this conflict of interests may be the underlying cause of the substantial reductions in milk protein content which occur at this time, and appear to be largely unavoidable. Currently, there is interest within the UK in feeding ruminally-protected protein prior to calving and its possible beneficial effect on milk protein content in the early stages of the subsequent lactation. Unfortunately, the results are equivocal and will remain so until it is accepted that the natural decline in milk protein content after calving may have both nutritional and endocrinological implications. Feeding the high yielding cow Both improved genetic and nutritional knowledge have contributed to the significant improvements in milk yield being achieved in many cows within the UK national herd. The current average of 5500 litres for all cows and heifers represents an increase of over 1000 litres in the last decade or so, yet an increasing number of cows are yielding in excess of 10,000 litres over a 305d lactation, and where farmers are attempting to dry off cows yielding 30+ litres/d prior to calving, the possibility of moving towards extended lactations is being seriously considered. In a recent review, Reynolds and Beever (1995) assessed the implications of feeding cows to achieve high yields of milk with acceptable levels of protein and fat. Taking into account the likely contribution from mobilised tissue, as discussed earlier, as well as the fact that peak intake is not usually achieved until some 3-4 weeks after attainment of peak milk yield, they concluded that many cows could be nutritionally compromised if total ration ME fell below 13.0 MJ/kg DM. This suggested that opportunities for including significant amounts of forage in the diet could be severely restricted, which would have major implications in relation to the integration of dairy farming into the wider aspects of UK agriculture. Furthermore, it emphasises the need for first class nutritional management, with the implications of any errors being potentially disastrous in terms of cow performance, longevity and profitability. Based on cows with milk yields of 60 litres/d, and fat and protein contents of 39.4 and 31.9g/kg respectively (current UK national averages) whilst losing 1.5 kg body weight/d, Reynolds and Beever (1995) estimated a minimum total dietary DM intake of 25.0 kg/d 7 Proc. Aust. Soc. Anim. Prod. I996 Vol. 21 would be required, with 80% of this derived from concentrates to provide a total ration ME density of 13.1 MJ/kg DM. If a reduction in ME density to 12.5 MJ/kg DM is unavoidable, this would necessitate an increased feed intake of 26.2 kg DM /d, whilst failure to achieve this would lead to either increased tissue mobilisation or impaired lactational performance. Based on experie