The influence of soil wetness on utilized output from grassland on commercial farms

Records of grassland productivity have been used for 93 farms in 1975–76 (a dry year) and 117 farms in 1977–78 (a wetter year). The samples were selected to contrast well-drained and poorly/ badly drained farms and were further subdivided into dairy and beef. The number of farms in each of the four subsamples for each year was between 19 and 32. In addition the number of days on which the soil was at meteorological field capacity (field capacity days) on each farm was calculated.
The mean utilized metabolizable energy (UME) output within the dairy sample was 45 GJ ha⁻¹ on both well-drained and poorly/badly drained farms. Within the beef sample the output was 41 GJ ha⁻¹ on well-drained farms and 37 GJ ha⁻¹ on the poorly/badly drained farms.
The correlation between fertilizer N and UME output was stronger on well-drained farms than on the poorly/badly drained farms in the wetter year (r = 0.69 v. 0.16 on dairy farms; r = 0.56 v. − 0.12 on beef farms). In the drier year the converse was found (r=0.15 v . 0.49 on dairy farms; 0.13 v. 0.44 on beef farms). The effect of field capacity days on output was inconsistent; only within the dairy sample in the wetter year did increased wetness appear to reduce output.
It is suggested that soil wetness may have only a small effect on utilized output from grassland on a whole-farm basis because (i) most farms have at least some well-drained land, (ii) much of the utilization damage occurs in relatively short periods in spring and autumn and (iii) despite having utilization problems, badly drained land is capable of growing large quantities of grass.     

Assessment of regional variation in climate on the management of dairy cow systems in Ireland using a simulation model

A dairy system simulator, Dairy_sim, was designed to assess the interactions between climate and management in spring‐calving milk production systems based on the grazing of grass pastures. The simulator comprises three main components: a grass herbage growth model, an intake and grazing behaviour model, and a nutrient demand model. The simulator was initially parameterized using the Irish National Dairy Blueprint. Sensitivity analysis indicated that the simulator was most sensitive to stocking rate, milk output per cow and nitrogen fertilizer inputs, but less sensitive to other variables. Field data from four grazing systems were used to test the simulator and it was concluded that Dairy_sim was suitable for evaluating the interaction of climate and management for rotational grazing dairy systems based on perennial ryegrass pastures with Friesian cows. The simulator, Dairy_sim, was then used to evaluate the effects of the regional climates of Ireland on system management. The results indicated that, between regions, herbage production at the same input of nitrogen may vary proportionally by 0·10 and that the length of the grazing season may vary by 0·25. It was concluded that the simulator could be a useful tool for developing region‐specific dairy production blueprints.     

Nitrate leaching from cut grassland as affected by the substitution of slurry with nitrogen mineral fertilizer on two soil types

A field experiment was conducted to find out whether there is any difference in risk of N leaching to groundwater when cattle slurry and/or mineral fertilizer-N was applied to cut grassland. The experiment was carried out over two consecutive years on two sites (one with a relatively wet sandy soil and one with a relatively dry sandy soil). Treatments were mineral fertilizer-N at annual rates of 0–510 kg N ha⁻¹ year⁻¹ and combinations of sod-injected cattle slurry (85, 170, 250 and 335 kg N ha⁻¹ year⁻¹) and mineral fertilizer-N (289, 238, 190 and 139 kg N ha⁻¹ year⁻¹). Yield responses indicated that in the short run, 0·44–0·88 (average 0·60) of the slurry-N was as available as mineral fertilizer-N. The total N input from mineral fertilizer and slurry was a worse predictor of nitrate leaching (     0·11) than the N surplus (i.e. the difference between total N input and harvested N) (     0·60). The effective N surplus, based on the difference between the summed inputs of the plant-available N and harvested N, proved to be the best indicator of leaching (     0·86). Annual N application rates of up to 340 kg plant-available N ha⁻¹ complied with the target nitrate concentration in groundwater of 11·3 mg N L⁻¹ set by the European Union in both years on the wet sandy soil, whereas on the dry sandy soil none of the treatments did.     

Herbage yield and animal production from grassland on three commercial dairy farms in South-East England

Yields of grass were measured on three dairy farms in 1981. Farm 1, on free-draining loamy soil over chalk, produced 10.3 t DM ha⁻¹ using 336 kg N ha⁻¹. Farm 2, on poorly-drained loam over Oxford clay, produced 12.8 t DM ha⁻¹ using 354 kg N ha⁻¹. Farm 3, on loam over clay with some free-draining sandy loam, produced 9.5 t DM ha⁻¹ with 169 kg N ha⁻¹. Allowing for differences in soil nitrogen and summer rainfall these yields were quite close to expectations from experimental results, despite considerable poaching damage to some swards.
Animal production records were kept and all livestock were weighed every 6 months, so as to calculate the utilized metabolizable energy (UME) output from grassland on each farm. These were 73, 62 and 59 GJ ha⁻¹ on farms 1, 2 and 3 respectively.
The UME output, expressed as a proportion of the yield of grass (converted to ME), was taken as the 'efficiency of utilization' of grass. This was 70% on farm 1,44% on farm 2 and 58% on farm 3. Results are presented separately for the summer and winter 6 months. Utilization by grazing appeared to be markedly reduced under very wet soil conditions. Major losses occurred in the conservation of grass.
The results begin to establish a valuable base of data from commercial farms.     

The effect of calving date and stocking rate on the performance of spring-calving dairy cows

A 3-year experiment on milk production systems was set up in 1989 to investigate the effect of calving date and stocking rate on the performance of spring-calving dairy cows. An early-calving herd (System A) with a mean calving date of 23 January and stocked at 2·9 cows per hectare was compared with two later calving herds (Systems B and C) with a mean calving date of 15 March. System B had a similar stocking rate to System A (2·9 cows ha⁻¹), while System C had a stocking rate of 2·6 cows ha⁻¹. The average lactation yields (kg) over the three years were as follows: 5872, 5444 and 5584 (milk)210, 204 and 215 (fat), 187, 184 and 189 (protein) and 261, 245 and 250 (lactose) for Systems A, B and C respectively. The average milk composition (gkg⁻¹) was: 36·0, 37·6 and 38·7 (fat), 31·9, 33·7 and 33·8 (protein) and 44·5. 45·1 and 44·8 (lactose) for Systems A, B and C respectively. Delaying calving date to coincide with the beginning of the grass-growing season (System B compared with System A) reduced milk yield per cow significantly (P < 0·05) in all three years of the experiment. However, fat and protein concentration were increased, resulting in no significant difference in the yield of fat or protein per cow. Reducing the stocking rate from 2·9 cows per hectare to 2·6 cows per hectare for cows calving in mid-March (System C compared with System B) increased milk yield per cow significantly (P < 0·05) in only one of the three years (1990). Milk fat content was increased significantly in 1990. Stocking rate had no other     

Accounting for residual effects of previously applied nitrogen fertilizer on intensively managed grasslands

Only 0·20–0·70 of the fertilizer-nitrogen (N) applied to grassland is taken up in herbage in the harvest directly following application. Residual effects at subsequent harvests can be large but are poorly quantified, and rarely taken into account in current management practices. An increased understanding of N-use efficiency per harvest can improve operational management. This study systematically assessed the residual effects of previously applied N fertilizer on N uptake, dry matter (DM) yield and soil mineral-N (SMN) during the whole of the growing season. It is based on field experiments conducted on peat and mineral soils in 1991–1994. Statistical models were derived for SMN, N uptake and DM yield as a function of previously and freshly applied N fertilizer. There were clear residual effects of previously applied N in later cuts. They were relatively greater at higher levels of N fertilizer. On peat soils, 0·15–0·25 of the N applied was recovered as SMN. On mineral soils the proportion was maximally 0·08. There was a clear relationship between SMN and N uptake in the subsequent cut on mineral soils but not on peat soils. The value of SMN as a tool to adjust fertilizer-N application rates was hence found to be limited. There were clear relationships between the amount of previously applied N and the N uptake in subsequent cuts, on both soil types and over the whole of the growing season. It was concluded that the total amount of previously applied N is a useful indicator for adjusting N-fertilizer application rates.     

Assessment of contrasting perennial ryegrasses, with and without white clover, under continuous sheep stocking in the uplands. 3. Herbage production, quality and intake

Herbage characteristics were monitored over 3 years (1985–1987) in three perennial ryegrass ( Lolium perenne L.) varieties as grass/only (200 kg N ha⁻¹) and grass/clover ( Trifolium repens L.) (75 kg N ha⁻¹) swards when continuously stocked with sheep. Although mean total annual herbage production was similar from the grass varieties, growth of Aurora, a very early-flowering variety, was higher than that of Aberystwyth S23 and Meltra (tetraploid) late-flowering types in spring. Herbage production from grass-only swards was 15% higher than from grass/clover swards.
In vitro organic matter digestibility (OMD) of Aurora and Meltra was 34 g kg ⁻¹ and 26 g kg ⁻¹ higher ( P < 0·01) than that of S23. Averaged over varieties, the OMD of grass/clover was 25 g kg ⁻¹ higher ( P < 0·05) than grass-only during the post-weaning period. Individual lamb digestible organic matter intake (DOMI) was 22% higher ( P < 0·05) on grass/clover than on grass-only during this period. Water-soluble carbohydrates content of Aurora was 42% higher ( P < 0·001) than that of both Meltra and S23.
Herbage characteristics were related to lamb performance reported previously. Lamb production was positively correlated with DOMI. However, the magnitude of differences in lamb performance between treatments was much greater than the extent of differences detected in herbage characteristics.     

Barley/pea mixtures as cover crops for grass re-seeds

Substantial annual fluctuations in the performance of arable silage mixtures, comprising Minerva forage pea and Goldmarker barley, were observed from ten annual sowings between 1977 and 1986, Total dry matter (DM) yield varied between 2·5 and 11·5 t ha⁻¹. Dry matter content also ranged widely from 10·3% to 20·0% and there was a large variation in the proportion of peas in the DM (0·44 to 0·90). Almost 80% of this yield variation was due to the fluctuation in pea performance and over 60% of all yield variation was related to the variation in sunshine hours and air temperature.
In 1983 and 1984 differences in seed rate of between 20 and 140 kg ha⁻¹ of Minerva and 80 and 160 kg ha⁻¹ of Goldmarker influenced the forage composition at harvest, with the pea component capable of outgrowing and suppressing the other species when sown at over 80 kg ha⁻¹. Very high yields of peas were associated with increasing stress on the undersown grass re-seed and were considered undesirable.
It was concluded that sowing rates of between 120 and 160 kg ha⁻¹ for barley and a maximum of 60 kg ha⁻¹ for peas would, in most years, provide the best compromise between attaining good arable silage yields and avoiding excessive dangers of damage of undersown grass re-seeds.     

Short-term leaf surface adhesion of heavy metals following application of sewage sludge to grassland

In the short term, surface-applied, digested, liquid sewage sludge may remain on grass leaf surfaces, leading to enhanced direct intake of potentially toxic elements by stock. The purpose of this field experiment was to investigate metal adhesion to herbage from liquid sewage sludge applications. Liquid digested sludge was applied at 0, 55 or 110 m³ ha⁻¹ to an Agrostis capillaris and Holcus lanatus sward that had been cut to either 4 cm (short) or 13 cm (long). The application rate of metals was relatively low. One day after application (day 1), 8–13% of the sludge solids applied had adhered to the short grass treatment, accounting for 35–62% of the dry-matter (DM) yield. On the long grass, 20–42% of the sludge dry solids adhered, accounting for 46–47% of the herbage DM yield. The content of all heavy metals in and on the grass herbage (Cu, Fe, Ni, Pb, Zn, Cr, Cd) was significantly increased ( P < 0·05) at day 1. Cu concentrations in and on the grass decreased to 25 mg kg⁻¹ after 16–29 d, Fe took 33–45 d to decline to 1000 mg kg⁻¹ whereas Pb took 12–18 d to decline to 30 mg kg⁻¹. It took 36 d for Cd, 43 d for Fe, Cu and Ni, and 57 d for Zn and Pb to be not significantly ( P < 0·05) different from the control. Grass growth was the most significant factor( P < 0·05 for all treatments and metals) influencing the decline in grass metal concentration, explaining 65–96% of the variation. The cutting treatment, sludge application rate and metal species also significantly affected the extent and rate of reduction in metal contamination over time.     

Sward dynamics of a smooth-stalked meadowgrass dominantwhite clover sward rotationally grazed by cattle and/or sheep

This experiment was carried out to study the responses of sward components (particularly white clover, Trifolium repens ) to grazing management in a natural sward dominated by smooth-stalked meadowgrass ( Poa pratensis ) syn. Kentucky bluegrass. Treatments during two grazing seasons (1989–90) were: cattle grazing alone (C); cattle grazing followed by topping (CT); cattle grazing followed by sheep grazing (CS); and sheep grazing alone (S). Mean target pre- and post-grazing herbage masses were 2200 and 1100 kg DM ha⁻¹, estimated by single-probe electronic capacitance meter. Sward component dynamics were monitored using turf dissections, marked white clover stolons, and ring-toss white clover leaf counts. Component and sward data for the C, CT, CS and S treatments respectively, were: number of white clover leaves m⁻², 1295, 1384, 1408, 900 (s.e. ± 108); number of leaves per growing point, 3·2, 3·4, 3·0, 2·8 (s.e. ± 0·2); herbage accumulation (t DM ha⁻¹), 5·16, 5·02, 5·87, 8·28 (s.e. ±0 08); rejected herbage (% pasture area) 39·7, 7·7, 16·0, 0 (s.e. ± 75); and annual net herbage production (t DM ha⁻¹) 3·39, 4·35, 4·99, 8·28 (s.e. ± 0.07). Swards grazed by sheep alone contained less white clover, but regrew quicker and produced more herbage than other treatments. Close topping or grazing by sheep following dairy cattle grazing decreased sward rejection by cattle. These treatments maintained more of the pasture in better condition for subsequent cattle grazing, resulting in greater net herbage production than where no post-cattle grazing treatment was used.     

Silage and milk production: comparisons between hays of different digestibilities as silage supplements

In three separate feeding experiments using a total of thirty individually-housed Ayrshire cows three silages made from perennial ryegrass were given ad libitum together with supplements of four different hays in the long form. The in vitro D-values of the silages ranged from 0·298 to 0·283, and the hays from 0·280 to 0·200. The daily intake of hay DM varied from 0·2 to 4·2 kg per cow and was given either without or with a daily maximum of 2·2 kg concentrate DM containing 379–527 g CP per kg DM. On average, 1 kg hay DM decreased silage intake by 0·24 kg DM with a range of 0·21–1·20 kg. The hay supplements had only small and non-significant effects on total DM intake, milk yield and milk composition, but increased the daily intake of drinking water. In three behavioural studies, the eating and ruminating times expressed as min per kg DM did not differ significantly between the various supplement treatments. It is concluded that hay has only a marginal value as a supplement for grass silage, although the hay could serve as a useful 'buffer' feed if the amount of silage was limited.     

The effect of rate and timing of phosphate fertilizer on the yield and phosphate off take of grass grown for silage at moderate to high levels of soil phosphorus

A series of twenty-one trials was undertaken during 1985–1988 to investigate the effect on lowland silage crops of 0, 40 or 80 kg ha⁻¹ phosphate (P₂O₅) fertilizer applied as triple superphosphate (46% P₂O₅) in the autumn, spring or as a split application (spring and after first cut). All sites had moderate to high sodium bicarbonate-extractable soil phosphorus contents (18–34 mg P 1⁻¹ in air-dried soil).
Significant yield responses were obtained at eight of the forty-two individual cuts (two cuts per site). When meaned over all sites, spring or autumn applied phosphate increased dry matter yield at both first and second cut (mean total increase 0·32 t ha⁻¹), though the effect was significant only at the highest phosphate rate. The time of phosphate application had no overall effect on yield. Herbage P concentration and phosphate off take were significantly increased at both cuts by both rates of phosphate. Generally, the most recent application had a significantly greater effect than other timings. These findings indicate that DM yield responses to freshly available phosphate can occur on soils of moderate P status.     

Variations in the dietary cation–anion difference and the acid–base balance of dairy cows on a pasture-based diet in south-eastern Australia

The variation in the dietary cation–anion difference (DCAD) and the urinary pH of dairy cows was examined over the year 1996–97 in Victoria, south-eastern Australia. Mineral concentrations in the pasture and dairy cow milk production were also examined. Three farmlets (A, B and C) under different feeding and management systems were used for the purpose of the study. Feeding management was based on grazed grass with stocking rates of 1·4, 2·5 and 4·7 cows ha^–1 for farmlets A, B and C respectively. Cows on farmlets B and C received more supplementary feed than those on the A farmlet.
The urine of the cows in each herd was sampled for pH twice monthly, after morning milking. A sample of the feed on offer the previous day was collected and analysed for crude protein, in vitro dry-matter digestibility and macrominerals. Milk yields were recorded on the same day as urine sampling and weather data for the previous day were also collected.
Pasture cation–anion difference was not greatly influenced by stocking rate or associated management practices, although mineral concentrations in pasture did vary. Urine pH was unaffected by changes in climate, management strategies (e.g. stocking rate), season and stage of lactation. Moreover, urine pH was also unaffected by changes in DCAD until the DCAD declined below approximately +15 mequiv. 100 g^–1 for two consecutive sampling periods.
It is concluded that when this threshold for DCAD (+15 mequiv. 100 g^–1) is breached, even in late lactation, a decrease in urine pH occurs. In south-eastern Australia, the DCAD offered to non-lactating cows in the last 2 weeks of pregnancy, in spring-calving herds, on a pasture-based diet is nearly always above that regarded as optimum in other feeding systems.     

Establishment and production of common sainfoin (Onobrychis viciifolia Scop.) in the UK. 2. Effects of direct sowing and undersowing in spring barley on sainfoin and sainfoin-grass mixtures

The dry matter (DM) yield and herbage quality of swards of sainfoin ( Onobrychis viciifolia ), meadow fescue ( Festuca pratensis ,) and tetraploid perennial ryegrass ( Lolium perenne ) grown in monocultures and in four sainfoin:grass mixtures (0·33 sainfoin:0·66 meadow fescue, 0·66 sainfoin:0·33 meadow fescue, 0·33 sainfoin:0·66 perennial ryegrass and 0·66 sainfoin:0·33 perennial ryegrass), established by direct sowing or undersowing in spring barley, were investigated over 3 years in a field experiment in the UK. Direct sowing produced a mean yield across all species and mixtures of 1·8 t DM ha⁻¹ in the establishment year, whereas undersowing produced no measurable yield except for that of the spring barley. Undersowing reduced the yields of sainfoin and sainfoin-grass mixtures in the first full-harvest year but not in the second. The annual yield of a monoculture of sainfoin was 7·53 t DM ha⁻¹ and that of sainfoin-grass mixtures was 8·33 t DM ha⁻¹ averaged over 3 years. Both sainfoin and the sainfoin-grass mixtures had higher annual DM yields than the grass monocultures. The mixture of 0·66 sainfoin:0·33 meadow fescue gave the highest mean annual yield (9·07 t DM ha⁻¹) over the 3 years. There was a higher proportion of sainfoin maintained in mixtures with perennial ryegrass than with meadow fescue. The proportion of sainfoin in sainfoin–meadow fescue mixtures declined from 0·62 in the first year to 0·32 in the third year, whereas the proportion in sainfoin–perennial ryegrass increased from 0·48 in the first year to 0·67 in the second year and remained stable in the third year.     

Annual cool-season legumes for forage production in mild winter areas

The forage potential of several annual winter legumes – crimson clover ( Trifolium incarnatum L.), faba beans ( Vicia faba L.), forage peas ( Pisum sativum L.), serradella ( Ornithopus sativus Brot.) and vetches ( Vicia sativa L. and Vicia villosa Roth.) – was evaluated over two growing seasons in the wet and mild winter areas of Galicia (north-west Spain) with a moderately acid soil, at the two normal harvesting dates in the region, April and May.
The results indicate that, in April, serradella and crimson clover (in the warmer location), can produce dry-matter (DM) yields of 4–5 t ha⁻¹, which is similar to other winter forages in the same area, with an average growing season of 212–237 d, corresponding to about 339–420 accumulated degree–days above 10°C (GDD₁₀). In May, mainly in the warmer location, fava beans, peas and vetches yielded on average 4·7–8·5 t ha⁻¹ DM, with an average growing season of 166–206 d but with only 233–278 GDD₁₀ units. Faba bean, with an average of 8·5 t ha⁻¹ DM in the May harvest, was the highest producing forage legume. The results also suggest that some cool-season legumes could fit into double-cropping systems based on summer crops, such as maize, because of their forage yields and nutritive value. However, in order to produce consistent DM yields, the date of sowing is crucial; they must either be sown much earlier than the dates in this study or the harvests must be delayed until the end of April.     

Effectiveness of cattle slurry as a sulphur source for grass cut for silage

Three trials were carried out in separate years during 1987–89 to investigate the effectiveness of cattle slurry as a source of sulphur (S) for grass cut four times per year for silage. They were located on different areas of the same field in Dorset, UK, in an area known to be sulphur deficient for intensive grass production. Yields of dry matter and S off takes in response to sulphur in slurry were investigated and compared with those from gypsum treatments supplying 0–100 or 0–75kg S ha⁻¹. Slurry was applied at two times of year, November or February, and at two rates, either alone or in combination with the two lowest rates of gypsum.
Annual increases in yield from the slurry applied in February compared with the yield when no slurry was applied were significantly related ( r ²= 0·96) to the rate of slurry sulphur applied. For applications in November, the relationship was not significant. Sulphur in slurry increased annual dry-matter yield with an average efficiency of 55% when compared with the sulphur in gypsum. The annual apparent recovery in the herbage of S from slurry was 21·8%, compared with 44·7% from the two lowest rates of gypsum.
Cattle slurry, in these and other trials in southwest England, contained on average 0·35 kg S m⁻³ and, at rates currently applied in farming practice or likely to be applied in the future, would be insufficient to satisfy annual requirements for silage in sulphur-deficient areas.     

Influence of some sward characteristics on the consumption of irrigated pastures grazed by lactating dairy cattle

Pasture variables measured in eight experiments at Kyabram, Victoria were used to quantify factors influencing the consumption of irrigated pastures grazed by dairy cows. In addition, whether intake prediction equations based on sward characteristics could be developed and applied in the field was also considered.
Herbage intake (DMI) was positively related to herbage allowance (HA) in all cases; however, the increase in daily DMI achieved for every additional kg herbage offered to a cow ranged from 015 to 0·35 kg. Herbage intake was also consistently related to residual herbage (RH). For every additional tonne herbage dry matter (DM) ha⁻¹ remaining after grazing, each cow consumed increased daily amounts of herbage ranging from 3·2 to 5·1 kg DM. However, of the other pasture variables, herbage mass (HM), pasture type (P_T) and digestibility of the herbage DM on offer (DMD) influenced herbage intake less consistently.
When the data from all experiments were included together, DMI was influenced by sward characteristics in the following manner: where LW = live weight of the cow (kg) and DE = duration of experiment (d).
When RH was considered instead of HA, only 61% of the variability in intake was accounted for compared to 80% when HA was considered:
In both of these equations, two variables other than the sward characteristics were significant; these were LW and DE. The significance of these variables in these relationships are discussed together with the possibility of using them for predicting DMI in other environments. However, it is concluded that they do not have general applicability and it is suggested that they may be useful only in the environment in which they were generated.     

Silage and milk production: comparisons between unwilted and wilted grass silages made with different additives

Four grass silages, all made in mid-July from second-harvest perennial ryegrass swards, were compared in a 16-week feeding experiment with twelve Ayrshire cows. Two silages were unwilted and two wilted. All the silages received formic acid ('Add-F') at the rate of 3 litres t^-1 either with formalin at the rate of 1 litre t^-1 or without formalin. The unwilted and wilted silages had mean dry matter (DM) concentrations of 200 and 243 g kg^-1, and in vitro D-values of 0·293 and 0·272 respectively. The silages were offered ad libitum plus 6 kg concentrates per cow per day. The daily intakes of unwilted and wilted silage DM were 10·2 and 9·2 kg per cow respectively on the formic acid treatment, and 10·2 and 9·2 kg on the formic acid + formalin treatment. The mean daily milk yield on the unwilted silage treatments was 19·2 kg per cow which was significantly higher than the yield of 17·2 kg per cow on the wilted silage treatments. The formalin had no significant effect on milk yield. The four silage treatments had small and non-significant effects on milk composition. It is concluded that the unwilted silages, which had excellent fermentation characteristics, were superior to the wilted silages as a feed for dairy cows.     

Establishment and production of common sainfoin (Onobrychis viciifolia Scop.) in the UK. 1. Effects of sowing date and autumn management on establishment and yield

The effects of sowing date and autumn management of sainfoin ( Onobrychis viciifolia Scop.) were investigated over 3 years in the UK. Replicated plots were sown between April and September in 2003 and 2004. Autumn management treatments were early and late cutting carried out in the establishment year and in subsequent years. Dry matter (DM) yields were measured over 3 years. One harvest was taken from April to July sowings in the establishment year and three harvests in each of the following years. DM yields from sowings in April and May were 2·8 and 3·3 t DM ha⁻¹, respectively, in the establishment year, which were higher ( P  <   0·001) than from sowings in June and July. Sowings from April to July yielded 10·9–12·5 t DM ha⁻¹ in the first full-harvest year, and 9·5–11·5 t DM ha⁻¹ in the following year. Sowings in August and September only gave 5 t DM ha⁻¹ year⁻¹. Early-autumn cutting of an established sward reduced yields of sainfoin at the second harvest in the first and second full-harvest years. Sowing in May had the lowest proportion of weed species (0·06) in the establishment year, and sowing in July had the highest (0·53). Crude protein concentration increased as the seasons progressed from 149·8 to 230·1 g kg⁻¹ DM.     

A plate meter inadequately estimated herbage mass in a semi-arid grassland

A resting plate meter was tested in the laboratory and on a field site to determine its effectiveness for estimating peak herbage mass on semi-arid grasslands. In laboratory tests, data from the plate meter closely predicted the herbage mass of four perennial bunch grasses (thirteen of sixteen tests with r ² ≥ 0·90) but the closeness of the prediction varied with the pressure of the plate. Field tests took place on a southern Arizona, USA semi-arid grassland in 2005, 2006, 2007 and 2008. Twenty teams of two or three people sampled the site; five teams in each year measured eight to thirteen plate heights and clipped the vegetation beneath. Consistent with a need for yearly calibration, a significantly different ( P  <   0·05) linear relationship was detected between 3 of the 4 years which was associated with differences in average herbage mass of dry matter (DM) (1525 kg ha⁻¹ for 2005, 2093 kg ha⁻¹ for 2006, 1338 kg ha⁻¹ for 2007 and 1370 kg ha⁻¹ for 2008). Plate height poorly predicted herbage mass within years ( r ²  = 0·21, 0·51, 0·49 and 0·41 respectively) with plate heights explaining half or less of the variability in field herbage mass and having a mean prediction error of 466 kg DM ha⁻¹. The plate meter technique had limited potential for estimating peak herbage mass in semi-arid grasslands.