In 1977, a four-course rotation was set up at Ropsley (UK) to study crop response to eight rates of nitrogen application (35–265 kg ha−1). This rotation continued until 1990 when continuous winter wheat was introduced. Results from 1978 to 1990 provide an opportunity to study the initial phase of cumulative effects from different rates of N fertilizer application on the recovery of N by cereals and the retention of N in the soil.

From 1978 to 1990, considerable variation in the recovery of nitrogen by winter wheat was observed. Neither rainfall nor drainage, as indicators of possible denitrification or leaching losses, provided a useful explanation for this, possibly because of the relatively dry conditions prevailing after spring fertilizer application. There was no evidence of increased soil N fertility, beyond single year residues, as a result of large N applications over the 13-year period.

In order to achieve the economic optimum grain yield, it was necessary to use N applications which produced inefficient recovery of N. Thus, greater return of N in crop residues and immobilization at relatively large N applications (>150 kg ha−1) contributed to an observed build-up in soil organic N over the period of study. Plots receiving, on average, 265 kg ha−1 appeared to gain c. 250 kg ha−1 N over control plots (35 kg ha−1) after 13 years of N application. Reducing the N application rate from the economic optimum to a more biologically efficient N rate (156 kg ha−1) was calculated to result in an average yield loss of 0·305 t ha−1 and cause an estimated £17 ha−1 loss in profit.

The crop–environment resource synthesis model for wheat, CERES–Wheat, was used to simulate yields from 1985 to 1993 at Ludhiana, India. The simulated anthesis and physiological maturity dates, grain and total biomass yields of wheat were compared with actual observations for the commonly grown cultivar, HD–2329. The simulated and actual dates of phenological events showed deviations from only −9 to +6 days for anthesis and −6 to +3 days for physiological maturity of the crop. The model estimated the kernel weight within 88–113% (mean 100%) of the actual kernel weights. The model predicted the grain yields from 80 to 115% (mean 97·5%) of the observed grain yield. Biomass yields were predicted from 93 to 128% (mean 110·5%) of the observed yields. The results obtained with the model for the eight crop seasons demonstrated satisfactory predictions of phenology, growth and yield of wheat. However, the biomass simulations indicated the need for further examination of the factors controlling the partitioning of photosynthates during crop growth. The results of this study reveal that the calibrated CERES–Wheat model can be used for the prediction of wheat growth and yield in the central irrigated plains of the Indian Punjab.

The effects of contrasting management systems either of infrequent rotational grazing by town milk supply dairy cattle, or of frequent defoliation by continuously grazing sheep and beef cattle, on the morphology of independent plants and populations of ‘Grasslands Roa’ tall fescue in mixed pastures, were measured over 1 year (1992/93) in New Zealand. Volunteer perennial ryegrass plants were also measured for comparison.

While both species exhibited a similar pattern of clonal growth, tall fescue developed more plants of higher branching complexity than perennial ryegrass, chiefly through maintaining more connective stems, as herbage production was confined to the three youngest branching orders in both species. Greater resistance to microbial degradation of old stems through poorer quality organic matter (wide C[ratio ]N ratio) compared to perennial ryegrass may be responsible for the greater complexity of fescue plants. In addition, tillering rates in tall fescue were three times lower which was offset by greater longevity and size of leaves and tillers, compared to perennial ryegrass. As a result, seasonal fluctuation in the distribution of plants among the various branching orders in tall fescue was small, producing a more stable population relative to the distinct seasonal changes in the population of perennial ryegrass plants. Grazing management had no effect on the seasonal population structure in either species.

Differences in plant structure due to grazing management were small, with only slightly more tillers on sheep-grazed than on cattle-grazed tall fescue plants. The major effect of grazing management was on dry weight or size of plant components. Cattle-grazed tall fescue plants were 120% heavier, with greater numbers and lengths of stolons and flower heads than those under sheep grazing. For the volunteer perennial ryegrass, the difference was only 65%, possibly due to competition from the more vigorous tall fescue under rotational cattle grazing.

Both species produced stolons throughout the year, although these were primarily associated with reproductive growth in spring. In tall fescue, an additional distinction was made between stolon and rhizome, the latter occurring mainly in the summer–autumn. Their possible functions in plant growth are discussed.

The water-use model (watyield) was used to estimate a number of variables which were then related to the actual yields recorded in long term sugarbeet experiments carried out at several sites in England over the last 30–40 years. Nine functions were developed which are based on yield/water-input, yield/crop evaporation and (the equivalent of) relative yield/relative crop evaporation relationships. In all cases, highly significant linear correlations were obtained but the levels of precision, and degree of site specificity, varied. For irrigation planning and water allocation, three approaches are recommended with increasing sophistication and, in general, precision. These are (1) relating yield increases to the estimated irrigation need; (2) relating actual yields to crop evaporation totals; and (3) determining actual yields for rainfed and irrigated crops from the calculated seasonal totals of the ‘effective’ solar radiation intercepted by the crop canopy. These values can be estimated with the model using standard weather and soil physical data appropriate to the locality and the crop.

Experiments using cultivars with differing degrees of striga resistance were conducted at two sites at Kamboinse in 1988 and at two locations (Kamboinse and Kouare) in 1989 in the Sudan-Savannah region of Burkina Faso. At each site, striga-free (SFP) and striga-infested plots (SIP) were selected. Two factors, location and genotype, were found to be associated additively with yield losses in soils infested by striga. The location effect was probably due to lower soil fertility in the SIP than the SFP plots under farming conditions. Yield losses in SIP relative to SFP ranged from 3·1%, at the experimental station, to 44·2% under farmers' field conditions. The genotype effect was evident at all locations. Depending on the susceptibility of the cultivars, it varied from 3·1 to 36·5% of the mean yield of SFP with an average of 31·4% in susceptible cultivars. The location effect was evident only at Kouare, where SIP plots were under continuous cultivation without appropriate soil fertility maintenance and/or restoration measures. This amounted to c. 19·4% of the mean yield in the SFP. To reduce yield losses in soils infested by striga, it appears to be necessary to grow high yielding, striga-resistant cultivars using agronomic practices which are known to improve soil fertility.

In 1992 and 1993, eight rates of 15N-labelled fertilizer (0–245 kg ha−1) were applied to winter wheat growing on the Ropsley long term field trial where eight different N amounts had been applied annually since 1978. The fate of the labelled N in the crop and topsoil (0–23 cm) was determined at harvest in the year of 15N application and in the first and second residual years.

By harvest in the second residual year, 60–77% of the original labelled application had been recovered in the crop and topsoil with 23–40% lost. These losses virtually all occurred within the first two growing seasons; there was no significant loss during the second residual year. Significant changes in the 15N balance were observed at N applications in excess of the range 140–175 kg ha−1 which suggested a marked decrease in the efficiency of N use and an increase in residual labile N in the soil at harvest. At low N applications (<175 kg ha−1), a positive added nitrogen interaction (ANI) was observed: 40–50% of this was a residual ANI due to the short or long term effect of applying N fertilizer, and the remainder was probably an apparent ANI due to pool substitution in the immobilization process. At large N applications (>175 kg ha−1) a negative ANI was observed: large N applications resulted in a net suppression of soil N uptake due to substitution by fertilizer for a limited plant N demand.

Rice, a cash crop, was grown in the alluvial valley bottom (fadama) of an inland valley region of central Nigeria. The main staple crop, maize, was grown on the adjacent upland areas. Both oxen and manual labour were used for initial cultivation (ploughing and harrowing for rice and ridging for maize). Times spent were 94·3 and 315·2 h/ha respectively for rice and 28·2 and 65·5 h/ha for maize. Plots cultivated by animal traction (AT) produced more weeds and required more time for weeding than manually cultivated ones. Thus although animal traction saved time at the most critical time of year, it did not save time overall. Total time values were 1045 and 1064 h/ha for rice using animal traction and manual cultivation respectively. Corresponding values for maize were 654 and 484 h/ha. Type of cultivation had no significant effect (P > 0·05) on yields of crops (4·50 and 4·55 t/ha, AT/manual, for rice and 1·60 and 1·83 t/ha for maize). Not weeding the plots reduced rice yields to 2·78 and 2·83 t/ha for ox and manual cultivation respectively and to virtually zero for maize.

Rice was grown in an inland valley (fadama) region of central Nigeria. Plots were cultivated using oxen either in the dry season or at the beginning of the wet season. Around 25% more time was required for dry season cultivation but this enabled an average saving of 53·5 h/ha during the most critical time of the year at the beginning of the wet season. Total time for all operations during the year was similar (2075 and 2150 h/ha for dry and wet season respectively) of which most was spent on weeding (1388 and 1527 h/ha). Weeding time could be reduced by the application of pre-emergence herbicide to 1042 and 1247 h/ha for dry and wet season cultivation respectively. Grain yields were 4·0, 4·8, 4·2 and 4·6 t dry matter/ha for dry season/no herbicide, dry season/herbicide, wet season/no herbicide, and wet season/herbicide respectively. Similarly, straw yields were 4·8, 7·0, 5·7 and 7·4 t/ha. None of the differences in yield was statistically significant.

Maize was grown in the upland areas surrounding an inland valley in central Nigeria in a randomized block experiment using six cultivation techniques (manual cultivation with a hand hoe (MC); ploughing in both directions to throw up a ridge (DPL); single ploughing with the seed placed on the ridge (SPL); ridging with a wooden, single tine, locally made ‘bush’ plough (BPL); single ploughing with the seed placed in the furrow (FPL) and ridging with a conventional ridger (RID) with or without pre-emergence herbicide (PEH) with two replicates. Initial cultivation times ranged from 29 to 70 h/ha (BPL<RID<SPL<FPL<DPL<MC). Total weeding time ranged from 220 to 512 h/ha (MC<DPL<RID<FPL<BPL<SPL) with PEH and from 431 to 763 h/ha (MC<SPL<DPL<RID<FPL<BPL) without PEH. Ox cultivation techniques were associated with higher weeding times and larger weed burdens. Total times for all field operations were 568–758 h/ha (MC<FPL<DPL<BPL<SPL<RID) with PEH and 791–870 h/ha (BPL<MC<SPL<DPL<RID<FPL) without PEH. Thus, although ox cultivation saved time at the most critical time of year (cultivation and planting), it did not save time overall.

Amongst the ox cultivation techniques, work inputs were 6·3–34·1 MJ/ha (BPL<FPL<SPL<RID<DPL) and draught forces 387–1377N (BPL<DPL<FPL<SPL<RID).

Yields of maize cobs were 2·55–4·0 t/ha (SPL<FPL<MC<RID<DPL<BPL) with PEH and 1·65–3·55 (RID<BPL<FPL<DPL<MC<SPL) without PEH. Except for SPL, PEH was associated with increased yield especially when used with ox cultivation.

As regards crop yield, time inputs and work input, there was no advantage to be gained from using a separate (and expensive) ridger for maize compared with a locally made plough or the plough also used for the main cash crop (rice).

At 1992 prices, the cost of PEH was about the same per ha as the price of labour saved on weeding, but additional benefits associated with PEH use were the avoidance of mid-season labour bottlenecks and an overall increase in crop yield.

Thirty-two N'Dama heifers were offered ad libitum Andropogon hay plus 10·2 g/kg liveweight (LW) groundnut hay (GNH) (L) or 10·2 g/kg LW GNH and 3·9 g/kg LW groundnut cake (GNC) (H). After 4 weeks on the diets, half of each group were infected intradermally with Trypanosoma congolense clone (ITC 50) (LI and HI). Peak parasitaemia occurred 6–8 days after infection and started to decrease c. 56 days later. No differences in parasitaemia were observed between LI and HI animals. Packed cell volume (PCV) fell in all treatments (by 5·4, 13·8, 3·7 and 9·4 units after 49–63 days post-infection (p.i.) for the L, LI, H and HI groups respectively) and significant effects of infection and diet were observed. GNH and GNC intakes were maintained during the trial; however, infected animals had a decreased intake of Andropogon hay. LI animals lost significantly more weight during the experimental period than the non-infected controls (−71·4 v. −13·7 g/day). Meanwhile, HI animals gained less weight compared with the H group (52·2 v. 167·6 g/day). Weight losses appeared to be due to decreased food intake. In the period 54–68 days p.i., plasma concentrations of albumin were lower and plasma protein was higher in infected animals. Plasma cholesterol concentrations were also lower in infected animals 54–68 days p.i. Plasma urea concentrations were higher in supplemented animals but were not affected by infection. The results showed that animals on a higher plane of nutrition showed less severe clinical signs of infection. However, for all the parameters considered, the magnitude of the difference between groups on different diets was similar for both infected and control animals, suggesting that mechanisms of resistance were not affected by the planes of nutrition considered.

Milk yield was determined by the weigh-suckle-weigh method over 2 years (1983 and 1984 calvings), for a total of 305 purebred Hereford (H×H) and first-cross Brahman×Hereford (B×H), Simmental×Hereford (S×H) and Friesian×Hereford (F×H) cows grazing three pasture systems at Grafton, New South Wales, Australia. The age of the cows ranged from 6 to 11 years. The data were used to evaluate different estimates of milk yield and to examine the effects of milk yield on growth of calves up to weaning.

Of all the cow traits studied, average lactation milk yield (average of early, mid and late-lactation milk yields) had the highest correlation coefficient with calf 210-day weight (r=0·73) and pre-weaning average daily gain (ADGtotal, r=0·73), and explained >50% of the variation in the calf traits. However, milk yield was also moderately correlated with other cow traits (liveweight and body condition score). Thus, to predict calf performance, milk yield data may not be required if detailed data on other cow traits are available. This is supported by the finding that differences in the coefficients of determination (R2) between models for calf 210-day weight and ADGtotal which included average lactation milk yield and other cow traits (highest R2=69%) and models which included other cow traits but no milk yield estimate (highest R2=57%) were <13%.

Cow breed rankings for average lactation milk yield were similar to those for calf 210-day weight and ADGtotal. On high quality pasture, S×H and F×H cows produced the most milk (S×H, 7·5 kg/day; F×H, 8·3 kg/day; B×H, 5·7 kg/day; H×H, 5·5 kg/day) and weaned the heaviest calves (S×H, 255 kg; F×H, 252 kg; B×H, 215 kg; H×H, 217 kg), while on low quality pasture, B×H and F×H cows produced the most milk (B×H, 4·2 kg/day; F×H, 3·7 kg/day; S×H, 2·9 kg/day; H×H, 2·7 kg/day) but B×H cows weaned the heaviest calves (B×H, 180 kg; F×H, 168 kg; S×H, 159 kg; H×H, 124 kg).

Sixteen 2-year old female sheep were fitted with ruminal and duodenal cannulae at Johnstone Memorial Laboratory, Lincoln University during 1989–90. They were offered, at 2 hourly intervals, a pelleted concentrate diet (900 g/day) and chaffed lucerne hay (100 g/day). In a split-plot experiment they were infused, intraruminally and at four rates, with potassium (providing 16, 26, 36 or 46 g K/kg food DM/day) and magnesium (providing 1·3, 1·8, 2·3 or 3·1 g Mg/kg food DM/day) within a Latin square design and with the liquid and solid phase markers 51chromium EDTA and 141cerium chloride. Net absorption of Mg before and after the duodenum was estimated from dietary intake, duodenal flow and urinary and faecal excretion of Mg.

Increasing K intake resulted in a decline in net absorption of Mg from the entire digestive tract, supporting data in the literature. Increasing K intake from 16 to 46 g/kg DM decreased urinary Mg excretion by between 0·14 and 0·30 g/day, the extent of which was independent of the level of Mg intake. At high K intake Mg absorption from the rumen was reduced, the amount absorbed ranging from 0·07 g Mg/day at intakes of 1·3 g Mg/day and 46 g K/kg DM/day to 0·66 g Mg/day at intakes of 3·1 g Mg/day and 16 g K/kg DM/day. However, at high K intake, and when Mg absorption from the rumen was reduced, net Mg absorption from sites distal to the rumen was increased to an extent which suggested compensatory absorption. Increase in K intake was associated with a consistent reduction in plasma Mg concentration which was independent of Mg intake. Increases in Mg intake resulted in increases in Mg absorption and plasma Mg concentration at all rates of K intake in direct proportion to rate of intake. The reduction in Mg absorption from the rumen at high K intake was associated with an increase (0·3 units) in pH of rumen digesta.

Rumen and caecal digesta were collected, under anaesthetic, from eight sheep offered either hay, pelleted concentrate or pasture at the Johnston Memorial Laboratory, Lincoln University during 1991. Subsamples of digesta were incubated at 39°C for 1 h after adjustment of pH within the range 0·5–12 by the addition of H2SO4 or NaOH. The samples were centrifuged at 30000 g for 30 min and magnesium (Mg) concentration measured in the 30000 g supernatant fraction and in total digesta to assess Mg solubility. In rumen digesta, Mg solubility declined from 0·86 at pH 5 to 0·30 at pH 7 and differences in response between diets were small. Magnesium solubility in caecal digesta was generally higher than in ruminal digesta, and particularly at pH values >6. At pH 7 the difference was twofold. Moreover, differences were observed between diets in the rate of decline in solubility in caecal digesta with increasing pH. At pH 5, 0·90 of Mg from hay and concentrate diets was soluble compared with only 0·8 for pasture. At pH 7, Mg solubility in caecal digesta from hay and concentrate-fed animals was almost double that from pasture-fed animals (0·64 and 0·62 v. 0·36, respectively). The implications of the findings for Mg homoeostasis in ruminants are discussed.

This group, which is concerned with the applications of mathematics to agricultural science, is sponsored by the Biotechnology and Biological Sciences Research Council. It was formed in 1970, and has since met at approximately yearly intervals in London for one-day meetings. The twenty-ninth meeting of the group, chaired by Professor R. H. Ellis of the Department of Agriculture, University of Reading, was held in the Wellcome Meeting Room at the Royal Society, 6 Carlton House Terrace, London on Friday, 21 March 1997, when the following papers were read.