Effects of harvest date of timothy (Phleum pratense) on its nutritive value, and on the voluntary silage intake and liveweight gain of lambs

Herbage

A grass sward, dominated by timothy, was established as an undersown crop of barley at the Forage Research Centre in Umeå, Sweden (63°45′N, 20°17′E). The following spring the sward was fertilized with 90 kg N ha⁻¹. During the period of primary growth, herbages at three maturity stages – early (16 June), mid (20 June) and late (26 June) – were harvested from the same field. At the early stage no heads were visible, although they could be felt in about 0·33 of the plants. At the mid stage, heads were visible in about 0·50 of the plants. At the late harvest more than 5 cm of the peduncle was visible in about 0·50 of the plants. The harvesting details, and botanical and chemical composition of the herbages, are presented in Table 1. The herbages were wilted in the field to standardize the DM contents to a targeted range of 250–300 g kg⁻¹, precision-chopped (nominal chop length of 30 mm) and stored in bunker silos with the addition of 4 L Promyr^TM ton⁻¹ fresh weight of timothy. Promyr^TM (Perstorp Speciality Chemicals AB, Perstorp, Sweden) is a mixture of formic acid (450 g kg⁻¹), propionic acid (200 g kg⁻¹) and ammonia (60 g kg⁻¹). When the feeding experiment started the silages had been stored for more than 3 months.

Experimental design

The grass silages, produced from the three harvest dates, were fed ad libitum to thirty-nine ewe lambs in a changeover experiment. The lambs were allocated to blocks based on their live weight at housing: light (20·0–26·0 kg), medium (26·5–31·5 kg) and heavy (32·0–36·0 kg). The mean live weights (standard deviation of mean) of the lambs allocated to these blocks were 23·4 (1·88), 29·0 (1·75) and 33·4 (1·23) kg respectively. Each block was randomly divided into three groups with four or five lambs in each. When allocating groups, litter size and genotype were also taken into account. The changeover experiment had a Latin square design with three treatments (dates of harvest) and three periods in three squares. Initially, for a 2-week period, all lambs received silage (produced from a timothy-dominated sward) that was not one of the experimental silages. Each of the three following experimental periods consisted of 4 weeks: an adaptation first week, followed by 3 weeks in which measurements were made.

Lambs, feeding management and measurements

The lambs used were crossbred (on average 0·66 White Swedish Landrace: 0·33 Texel) ewe lambs. They were born indoors between 25 May and 12 June and had a mean birth weight of 4·3 kg. Of the thirty-nine lambs, seven were born as triplets, twenty-three were twins and nine were singles. They went to pasture with their dams at 2–3 weeks of age. Grazing continued until 24 September, when the lambs were housed. They were dewormed (fenbendazole; Axilur^®vet, Intervet, Stockholm, Sweden) and sheared the week after housing. Throughout the experiment the lambs were kept in groups in straw-bedded pens.

The experiment started on 10 October. The lambs then were on average 19 weeks old and their mean live weight was 32·4 (4·96) kg. During the experiment the lambs had free access to silage. A refusal margin of 0·10 was allowed. Silages and refusals were weighed for each pen once a day and the rations were adjusted once a week. Water and salt blocks were freely available in all pens. The recommended amount of minerals (Effekt Fårmineral^®; Lactamin, Kimstad, Sweden), which did not contain salt, was fed twice a week. To avoid differences in voluntary intake resulting from variations in the crude protein (CP) concentration of the silages, soya bean meal with 505 g CP kg⁻¹ DM and 129 g neutral-detergent fibre (NDF) kg⁻¹ DM was given to the lambs to bring the overall CP concentration of their diet to 160 g kg⁻¹ DM. The amount of soya bean meal was adjusted once a week, depending on the voluntary intake of the silage, and averaged 0·07 kg DM d⁻¹ for the lambs that were fed the mid-harvest silage. The late-harvest silage was on average supplemented with 0·12 kg DM d⁻¹ of soya bean meal. Lambs that were fed the early-harvest silage did not get any soya bean meal. There were no refusals of soya bean meal.

The lambs were weighed at the start of the experiment and after weeks 1 and 4 of each experimental period. All lambs, except the twelve lightest, were slaughtered after the experiment. A rough comparison of slaughter results was made, based on carcase weight (0·98 × warm carcase weight) and the last experimental weighing, which was registered 5 d before slaughter.

Chemical analyses of herbage and silage

At each harvest three samples of fresh herbage were cut (each from an area of about 50 cm in diameter) from a diagonal across the field, pooled and the botanical composition recorded. Samples of herbage for chemical analysis were taken from every load that was stored in the silos and pooled into three samples per silo that were analysed in terms of their DM content (AOAC, 1984; method 7.003) and concentrations of CP (AOAC, 1984; method 7.015), NDF (Chai and Udén, 1998) and water-soluble carbohydrates (WSC) (Larsson and Bengtsson, 1983), and to estimate the concentration of metabolizable energy (ME) (Lindgren and Lindberg, 1988).

During the feeding experiment samples of the silages offered and the refusals were taken daily and pooled within each experimental period. The pooled samples were analysed for DM content, the concentrations of CP, NDF and WSC and the estimated concentration of ME, using the same methods as for the fresh herbages. The silage samples were also analysed for concentrations of volatile fatty acids (VFAs) (Andersson and Hedlund, 1983), pH and ammonia-N. The volatile N fraction in the silage juice was distilled in a Kjeltech Autosystem 1030 (Tecator AB, Höganäs, Sweden). When calculating nutrient intake, the concentration of nutrients of the silage was corrected for the concentration in the refusals, based on the mean weight of refusals each week. Soya bean meal was sampled each day, pooled and analysed per experimental period in terms of DM content (AOAC, 1984; method 7.015), and CP (AOAC, 1984; method 7.015) and NDF concentrations (Chai and Udén, 1998).

In vitro gas incubations of silages

Representative samples for the in vitro studies were taken 4 months after ensiling from each bunker silo by drilling cores (diameter 45 mm) from top to bottom in a well-distributed horizontal grid. The six core samples were pooled and mixed. The pooled samples (c. 5 kg) were dried at 60°C for 20 h and milled in a hammer mill through a 1-mm screen (Kamas Kvarnindustrier AB, Malmö, Sweden). The in vitro degradation characteristics of the milled samples were then analysed using an automated in vitro gas recording system (Cone et al., 1996) as follows. Portions containing about 400 mg of organic matter (OM) were incubated at 39°C in 60 mL of buffered rumen fluid in 250 mL serum bottles (Schott, Mainz, Germany). The bottles were continuously agitated gently. After 72 h the samples were placed on ice to terminate the fermentation. Each in vitro analysis was carried out in duplicate, including portions of buffered, sample-free rumen fluid as blank controls, in three consecutive series. The gas produced was recorded every 12 min, corrected for gas produced in the blanks and adjusted to standard atmospheric pressure (1013·5 hPa) (Hetta et al., 2003). Rumen fluid was obtained from two dairy heifers (Swedish Red) fed a ration of timothy hay, by oesophageal sampling with a flexible probe. This sampling, as well as all animal experimental work at the department, had been approved by the local Animal Ethics Committee. The rumen fluid was sampled 2 h after the morning feed for each run, then mixed and filtered into a buffered mineral solution (Menke and Steingass, 1988) with a rumen fluid:buffer ratio (v/v) of 1:2. The microbial activity of the rumen fluid was checked in each batch by including a standard hay sample with a known profile of GP. The fermentation residue was filtered through fritted discs (porosity 2) under a light vacuum to remove the liquid fraction. The remaining residues were treated with a neutral-detergent solution using the oven method of Chai and Udén (1998). The samples were then repeatedly washed with de-ionized hot water and finally rinsed in acetone to remove all residual detergent. The crucibles were dried overnight at 105°C, weighed, incinerated at 550°C and then weighed again. The degradability of the OM and NDF was determined from the NDF residues.

Treatment of data and statistical analyses

The data of cumulative GP for each sample were fitted to a monophasic model, as described by Groot et al. (1996) (Eqn  1), using Tablecurve 2D (Ver. 5.0®; SPSS Inc., Chicago, IL, USA). The following sigmoid equation predicts the GP (mL gas g⁻¹ OM) at a given time t (h).

(1)

where A (mL gas g⁻¹ OM) is the asymptotic GP, B (h) is the time at which half of the gas has been produced and C determines the sharpness of the switching characteristics of the curve and is dimensionless. From the degradation parameters B and C the time of the maximum fractional GP rate (T_max, h) and the maximum fractional GP rate (R_max, h⁻¹) were calculated for each silage (Groot et al., 1996). Comparisons between the mean composition of the silages and the corresponding refusals and between degradation parameters for the different silages were made with one-way analysis of variance, and pair-wise comparisons were made with Tukey’s test.

The effects of treatment (harvest date) on voluntary intake, liveweight gain and feed conversion efficiency were explored by analysis of variance using SAS (Ver. 8.02; SAS Institute Inc., Cary, NC, USA). The model used was y = μ + ax + bz + cv + e, where μ is the overall mean, x is the effect of treatment, z is the effect of block (liveweight class) and v is the effect of time (experimental period).

Correlations (Pearson correlation) between voluntary intake of DM of silage and GP characteristics were calculated with treatment means as the experimental units. Correlations between voluntary intake of DM of silage and chemical characteristics of the silages were calculated utilizing group × period means as the experimental units.

Voluntary intake, liveweight gain and feed conversion efficiency

The harvest date had significant effects on both total voluntary intake of DM and DM intake of silage, as shown in Table 4, and this was also the case when intake was calculated on the basis of live weight (P < 0·05). The DM intake per kg liveweight (kg⁻¹ LW) by the lambs of the early-harvest silage was 30·9 g kg⁻¹ LW. The corresponding values for the medium- and late-harvests were 25·1 and 20·5 g kg⁻¹ LW respectively. The effect of date of harvest on liveweight gain and feed conversion efficiency was also significant (P < 0·001), as shown in Table 4.

The positive linear correlations between DM intake of silage and OM degradability, NDF degradability and the GP parameter A were over 0·95 (P = 0·08–0·15, n = 3). There was a negative correlation of −0·98 between DM intake of silage and T_max (P = 0·11, n = 3). The correlations were 0·70–0·75 between DM intake of silage and the concentration of WSC, CP and estimated ME in silage and, and −0·74 between DM intake of silage and NDF concentration in silage (P < 0·001, n = 9).

There was a significant effect (P < 0·001) of liveweight class on voluntary intake of DM, CP, NDF and estimated ME by lambs for both total intake and silage intake only (Table 4). The lightest liveweight class had the lowest daily voluntary intake per lamb. The daily DM intake of silage in that liveweight class was 0·90 kg, compared to the heavier lambs that consumed 1·10 kg. When calculated per kg LW, the heaviest lambs had the lowest DM intake of silage (23·7 g kg⁻¹ LW). The corresponding values for the medium and lightest lambs were 27·3 and 25·5 g kg⁻¹ LW respectively (effect of liveweight class, P < 0·05).

There was a significant effect (P < 0·001) of experimental period on the intake per kg LW, which was higher in the first period than in the two subsequent periods. Total DM intakes in the three periods were 31·0, 25·9 and 24·3 g kg⁻¹ LW in periods 1, 2 and 3. There was also an effect of period on liveweight gain (P < 0·01): the mean liveweight gains during the first, second and third periods were 144, 105 and 103 g d⁻¹ respectively.

The mean carcase weight of the lambs that were slaughtered at the end of the experiment was 19·9 (2·04) kg, with a dressing out percentage of 40 (2·1). There was no significant effect of harvest date on carcase weight.

Chemical composition of timothy and degradability characteristics of the silages

The harvest window for timothy is narrow, because of rapid maturation of the cell walls. This has been observed in previous studies of both pure and mixed timothy swards (Rinne and Nykänen, 2000; Gustavsson and Martinsson, 2004; Hetta et al., 2004). Although it was only 10 d between the earliest and the latest harvest in this study, it resulted in herbages with distinctly different nutritive values (Table 1).

In temperate grasses large proportions of nutrients are allocated to soluble fractions at early stages of maturity (Schofield and Pell, 1995). For silages made from these herbages it is important to use a technique that can record the degradation of soluble, as well as non-soluble nutrients, such as by the use of the GP technique (e.g. Beuvink and Spoelstra, 1994; Cone et al., 1999; Hetta et al., 2004). The results of the in vitro analyses (Table 3) show that, as the grass matures, the potential availability of substrate to the rumen microbes declines, as demonstrated by the reduction in the degradability of OM and NDF. The decline in degradability of the silages is probably partly caused by an increased proportion of stems in relation to leaves in the plants, as suggested by Rinne and Nykänen (2000).

Even though the plants decreased in degradability with date of harvest, the estimated rate of degradation (R_max) of the silages from grass harvested at all three stages of maturity was relatively high compared to other studies. The estimated rates in Table 3 are slightly higher than the corresponding rates recorded by Schofield and Pell (1995), and comparable to GP measurements of fresh perennial ryegrasses with high concentrations of WSC reported by Miller et al. (2001). These values indicate that the silages were readily degradable at all three stages of maturity. The differences in rates of degradation between treatments could, to some extent, be related to the differences in NDF concentrations. This conclusion is in line with the findings of Hetta et al. (2004) who found that the neutral-detergent-soluble fraction of timothy degrades three times faster compared to the NDF fraction, independent of the maturity of the grass. The importance of the ratio between the neutral-detergent soluble fraction and NDF is supported by Mertens (1993) who concluded that the degradation of forages is mainly limited by the concentration of cell walls (NDF) and secondarily by the quality of the cell walls (degradability of NDF). It is important to note that the in vitro analyses in this study were conducted on the silages offered to the lambs rather than the feed that the lambs actually consumed. Based on the higher concentrations of NDF in the refusals (Table 2), it is likely that the true diets were even more readily degradable than the values reported in Table 3.

Intake of silage by lambs may be affected by the concentration of fermentation products (e.g. Wilkins et al., 1971; Offer et al., 1998). The silages in this study were well preserved, with comparable concentrations of fermentation products, which increases the scope for correctly estimating the effects of herbage maturation without disturbance from confounding effects (Huhtanen et al., 2007). The silages were comparable to those examined in other studies of DM intake of silage by lambs in terms of pH and fermentation products (Offer et al., 1998; Fraser et al., 2000). The concentration of ammonia-N was somewhat elevated compared to Swedish standards for good-quality silage (maximum of 80 g ammonia-N kg⁻¹ N), probably because the silage additive Promyr™ itself contains some ammonia.

Voluntary intake, liveweight gain and feed conversion efficiency

One of the measurements most commonly used to estimate the physical fill of feeds for ruminants is the concentration of NDF (see review by Allen, 1996). Sheep have been reported to have a lower capacity to digest fibrous forages than cattle (Südekum et al., 1995) which may explain the relatively strong negative effect on DM intake of the silages with the higher concentrations of NDF (Table 4). The negative correlation between DM intake of silage and its NDF concentration was highly significant in this study. This finding is supported by a study where the correlations between in vitro degradation characteristics, chemical composition and DM intake for nine different silage diets were evaluated (M. Hetta, unpublished data). In that study the concentration of NDF was the most important variable in explaining the variation in DM intake of silage between diets. Together with live weight of the lambs, it explained more than 0·75 of the variation in intake of silage.

The lambs fed the later cut silages had similar liveweight gains to those observed in store lambs fed leafy perennial ryegrass silage in a study by Fitzgerald (1996), indicating that even the late-harvest timothy had a high nutritive value, as one could assume from the relatively high rate of degradation recorded in vitro. The rate of degradation by microbes, which has been used in many studies to explain variations in intake between feeds (e.g. Khazaal et al., 1995; Rodrigues et al., 2002), is likely to be an important factor in the physical regulation of DM intake of silages. The differences in rate of GP found between harvest dates in this study may, therefore, explain a large part of the effects on DM intake of the silages.

There could have been a risk of carry-over effects between periods in the experiment design – lambs fed the late-harvest silage in the first period and the early-harvest silage in the last period may have had an advantage over the other treatments in the last period, due to a potential effect of compensatory growth. This could have biased the results in favour of the early-harvest silage. This is unlikely because in the first period there was a significant effect of harvest date on liveweight gain (P < 0·05).

An important factor in changeover studies is to take into account the time needed for adaptation of the rumen microbes to new feeds. The results presented in this study are based on 3 weeks of measurements plus 1 week of adaptation in between, which may be too short a period if the treatments differ much in composition (Morris, 1999). Calculations were made using only the last 2 weeks in each experimental period (results not shown). This did not change the statistical interpretation of the treatment effects.

The growing lamb has high demands for metabolizable protein and energy (Hegarty et al., 1999) which may explain why the treatments had a strong effect on liveweight gain in this study. The higher DM intakes of silage from the early and medium-harvest silages supplied the lambs with more dietary protein due to their higher concentrations of CP. It is possible that there was also an increased utilization of microbial protein, due to the higher availability of structural carbohydrates (Stern et al., 1994) in the silage from the less mature herbages, as can be seen from the values of degradability of NDF in Table 3.

Calculated for different harvest dates, the results for feed conversion efficiency indicate that harvest date was likely to have a strong effect on the liveweight gain of lambs. The most likely explanation for the higher feed conversion efficiency of the early-harvest silage is the higher concentration of neutral-detergent solubles, which are almost completely digestible (Van Soest et al., 1994) and which will be utilized even at high rates of passage of digesta in the rumen.

Sheep and other small ruminants have the capacity to optimize the composition of their diet by selecting the most nutritious fractions of forages offered to them. Studies on the voluntary intake by sheep should, therefore, include quantification and analysis of refusals (Van Soest et al., 1994). The effect of diet selection is clearly demonstrated by the comparison between the feeds and refusals in this study (Table 2). The results show that the lambs were able to select the more nutritious components from all silages offered, and that their selection was based more on the concentration of NDF and estimated ME than on CP. The higher concentration of ammonia-N and differences in VFA concentrations in the refusals compared to the silages is probably due more to deterioration of the silages in the feed trough than avoidance by the lambs, even though ammonia-N is known to have a negative effect on the DM intake of silages by ruminants (e.g. Wilkins et al., 1971; Offer et al., 1998; Huhtanen et al., 2002). Furthermore, the significant effects of treatment (harvest date) on lamb performance indicate that, although the lambs were able to select the more nutritional fractions of their feed, their selectivity could not fully compensate for the differences in the nutritive value between the silages.

The values for voluntary intake relative to live weight observed in this study are comparable to those recorded in previous studies of lambs in northern Sweden. In a review of five experiments, Måntelius (2000) found that the daily DM intake of silages averaged 30 g kg⁻¹ LW. In this study the heaviest lambs had the highest daily intake but their intake per kg LW was lower than those of the lighter lambs, probably because the former were closer to their maximum live weight. This fact could also partly explain the lower DM intakes and liveweight gain in the later experimental periods. The effect of period on DM intake and liveweight gain may also depend on the changes in day length during the experiment. Studies have shown that shorter day length in winter is associated with reductions in both intake and liveweight gain in sheep (Forbes, 1982; Iason et al., 1995; Bernes, 1996).