Materials and methods 7.5 and other ingredients. Cows received 1 kg of

28 H . Valk et al. Livestock Production Science 63 2000 27 –38 Minson 1990. However, little is known about the and in August–September late summer of 1992 effects of reduced amounts of N fertilizer on VI and A and 1993 A . Before the start of each 92 93 animal performance of dairy cows during the grow- experiment 36 multiparous Holstein–Friesian ing season. Most experiments in literature were crossbred dairy cows in mid lactation were allocated carried out with sheep offering very mature grass. It to 12 blocks of three cows on the basis of calving is uncertain whether the results of those experiments date, milk yield and milk composition. Means of the can be extrapolated to dairy cows grazing grass in stage of lactation and milk production data of the 12 earlier stages. When grass is harvested on the same cows per treatment at the start of the experiments, day after regrowth, N fertilizer has no consistent are given in Table 1. Cows were housed in tie stalls effect on VI, indicating that the effect of N on VI is on rubber mats and had free access to water. Within strongly related to stage of maturity Minson, 1990. each block cows were allotted at random to one of Therefore, more information is needed about the the three treatments. Treatments were three rations effects of reducing N fertilizer on VI and milk containing fresh grass harvested daily and fertilized performance of dairy cows offered grass cut at with either 450 N , or 300 N , or 150 N 450 300 150 similar dry matter DM yields. kg N ha per year. Cows received daily 1 kg in the The objective of our experiments was to quantify early and 2 kg compound feed in the late summer the effects on VI and milk performance of dairy cows experiments which amount was offered in two equal fed indoors with grass grown at three different portions during milking. This compound feed con- annual levels of N application 450, 300 and 150 tained sugar beet pulp 20, citrus pulp 20, kg ha during two different periods spring early maize gluten feed 22.5, palm kernel expeller summer and late summer. In order to minimize the 15, soybean hulls 8 and other ingredients. effect of stage of maturity on VI and to connect with Faecal output was estimated by marker technique. the practical grazing situation, VI of grass is tried to Therefore, 4 g Cr 0 per kg was added to another 2 3 compare at DM yields between 1500 and 2000 kg ha compound feed Cr compound containing sugar for the three N treatments. beet pulp 15, maize gluten feed 20, palm kernel expeller 17.5, soybean hulls 14, lin- seed expeller 8, coconut expeller 8, wheat

2. Materials and methods 7.5 and other ingredients. Cows received 1 kg of

Cr compound per day in two equal portions after 2.1. Cows and treatments milking. Chemical composition and nutritive value of the commercial and Cr compound feeds did not Experiments were carried out in May–July differ between years. They contained per kg DM: 93 spring early summer of 1991 S and 1992 S and 101 g ash, 178 and 163 g CP, 50 and 46 g CF, 21 91 92 Table 1 Stage of lactation in days after parturition LS, milk yield and composition and live weight LW of 12 cows per ration N , N and 450 300 N at the start of the experiments carried out in spring of 1991 S and 1992 S and in late summer of 1992 A and 1993 A 150 91 92 92 93 Experiment S S A A 91 92 92 93 N N N N N N N N N N N N 450 300 150 450 300 150 450 300 150 450 300 150 LS d 133 122 136 173 175 157 180 177 170 163 196 222 Milk kg d 29.1 28.9 29.3 28.7 28.7 28.7 31.1 31.9 31.6 30.6 30.8 30.6 Fat g kg 46.5 47.1 46.4 45.7 45.7 45.6 40.0 40.2 39.8 40.7 40.6 40.3 Protein g kg 35.4 35.6 35.5 36.6 35.8 35.8 32.5 32.2 32.5 34.1 34.4 34.3 LW kg 621 648 621 594 621 649 595 605 595 604 623 617 H . Valk et al. Livestock Production Science 63 2000 27 –38 29 and 103 g starch, 347 and 383 g NDF, 97 and 70 g the third harvest of 150N plots was fertilized with an DVE Tamminga et al., 1994 and 1061 and 1032 extra supply of 40 kg ha. VEM Van Es, 1978 for the commercial and Cr compound, respectively. The experiments carried out 2.3. Sampling and measurements in spring early summer started on 13 May in S and 91 on 18 May in S and lasted for 10 weeks. The 92 Grass was harvested daily with a cyclo mower experiments in late summer started on 27 July in A 92 starting at 09.00h, analyzed for DM content by a and on 2 August in A and lasted for eight weeks. 93 rapid method described by Meijs 1981 and For all experiments an adaptation period of two weighed in 13 14 portions of 1.4 kg DM grass for weeks is included. each cow. After milking in the afternoon about eight grass portions were offered during the time between 2.2. Meteorological and agronomy aspects 16.00h and 21.00h. The remainder portions were stored at 38C and offered after the morning milking Rainfall was measured daily using a Lambrecht between 07.00h and 16.00h. After the adaptation rain gauge. Temperature was measured every 10 min period, intake was measured individually on four with thermometers consisting of platinum electrodes. consecutive days in each week by weighing grass Grass was grown on clay soils and contained 70 refusals once daily directly before milking in the Lolium perenne, 20 Poa trivialis and 5 Poa afternoon. On these days, grass samples were taken annua. About 30 ha was divided into 26, 22 and 20 which were oven dried at 708C and analyzed for numbered plots 300 m 3 12 m per plot each DM, ash, Kjeldahl N, crude fibre, sugar and neutral receiving an annual application of 150, 300 and 450 detergent fibre NDF according to methods de- kg N ha, respectively. In early spring these plots scribed by Van Vuuren et al. 1993. In-vitro organic received respectively 20, 50 and 65 kg N ha prior to matter digestibility in-vitro d was estimated the first cut. Each day from mid-April onwards, three OM according the method of Tilley and Terry as modi- grass plots starting with plot number 1 of each N fied by Van der Meer 1986. Samples of grass treatment were mown and directly harvested after refusals were collected daily per N treatment, which which each plot received an application of 30, 60 were analyzed for DM and ash content. At the start, and 90 kg ha for the second cut. Plots were mown at middle and end of the experiments, samples of the target yields between 1500 and 2000 kg DM ha. The compound feeds were taken and analyzed for the spring experiments started when the first plot of the same components as grass. In addition, crude fat and 150N regime second cut had reached that target starch were analyzed in the compound feeds accord- DM yield. Fertilization and mowing activities were ing the methods described in Van Vuuren et al. recorded on a grassland calendar. Third and fourth 1993. Samples of the Cr compound were taken cuts received the same amount of N as the second every week and analyzed for DM, ash and Cr see cut. The remaining harvest received 20, 40 and 60 kg determination method by Van Vuuren et al. 1993. N ha for the three N treatments, respectively. Grass The energy value of the feed stuffs was calculated as growth was based on weekly grass DM yields net energy for lactation and expressed in VEM units obtained from nine plots 3 per N level using an according to Van Es 1978. Protein value was Agria motor scythe Meijs, 1981. Due to the calculated as true protein digested in the small difference in cutting height between the Agria motor intestine DVE and the degraded protein balance in scythe and the cyclo mower 3 versus 6 cm cutting the rumen as OEB Tamminga et al., 1994 based on height, the nine plots were also cut with the Agria CP content and day of harvesting. just before and after the harvest of the whole plot by To estimate total apparent digestibility, faecal the cyclo mower. Data of grass growth were used to samples were taken individually from nine cows per perform the strategy of harvesting the swards at DM N treatment during six weeks in the spring early yields between 1500 and 2000 kg ha. summer and five weeks in the late summer experi- During a drought in 1992 A , all plots were 92 ments. During three consecutive days per week, each irrigated with 24 mm water in calendar week 27 and 30 H . Valk et al. Livestock Production Science 63 2000 27 –38 dung patch excreted between 06.30h to 15.00h was above. Statistical analysis was carried out using sampled and pooled to one sample per cow per Genstat Genstat 5 Committee, 1993. Treatment week. Samples were dried at 708C and analyzed for means were compared by Student’s t-test. In experi- DM, ash, NDF and Cr content and Kjeldahl N was ment A one cow was infected by a severe form of 93 analyzed in a fresh subsample. The fixed amount of mastitis and was assigned as a missing value in the Cr ingested each day and the Cr content in the faeces statistical analysis of the data. The effect of N assuming a Cr recovery of 94, were used to fertilizer on chemical composition and nutritive estimate daily faecal output per cow per week faecal value of grass was tested for significance using a output 5 ingested Cr faecal Cr content. linear regression model with N treatment as the only Cows were milked twice daily with milk yields variable. being recorded at each milking. Milk samples were taken during six consecutive milkings per week for fat and protein analyses, which were determined by

3. Results