ALBERT

Description: SUMMARYExperiments with winter wheat in 1972 and 1973 tested all combinations of ‘Nitro-Chalk’ ν. liquid N-fertilizer, 56 ν. 112 kg N/ha, 0 ν 5·6 1/ha of herbicide (2·8 kg acid equivalent/ha) and 0 ν 0·7 1/ha of mildew fungicide, all applied at growth stage 4–5 of the Feekes scale. The liquid fertilizer (26 % N) was a solution of ammonium nitrate and urea, the herbicide was a mixture of dichlorprop and MCPA and the mildew fungicide contained 75 % (w/v) of the active ingredient tridemorph.The herbicide and mildew fungicide were sprayed either alone or together and neither scorched the wheat leaves. Liquid N-fertilizer by itself slightly scorched the wheat leaves and scorch was increased by adding herbicide to it, but more by adding the fungicide and most by adding both; it was then severe, especially with 112 kg N/ha.Weed control after adding herbicide to the liquid fertilizer was at least as good as from herbicide sprayed alone.In July, foliar diseases were much more severe with 112 than with 56 kg N/ha, but effects of the other treatments, including fungicide, on foliar diseases, were then very small.With 56 kg N/ha, yields were slightly larger with ‘Nitro-Chalk’ alone than with the liquid N-fertilizer alone, but with 112 kg N/ha they were slightly larger with the liquid fertilizer; adding herbicide to the liquid fertilizer did not change these results. With either amount of N, adding mildew fungicide to the liquid fertilizer made it less good than ‘Nitro-Chalk’, presumably because of the damage from leaf scorch; adding both herbicide and fungicide to the liquid fertilizer increased the damage.

Description: SummaryMidas, a barley variety very susceptible to brown rust (Puccinia hordei), and two less susceptible varieties (Julia and Mazurka) were given three amounts of nitrogen (50, 100 or 150 kgN/ha) either at sowing (in March), in May, or half at each time. Benodanil sprays were applied to one half of each plot in June and again in July to control brown rust. Brown rust became more severe in 1973 than in 1974 and was most severe on Midas with 150kgN/ha. In 1973, there was most brown rust where all or part of the nitrogen was applied in May, but in 1974 there was least brown rust where all the nitrogen was applied in May, probably because dry weather limited its uptake.In 1973, when ample rain fell in spring and summer, grain yield was increased more by nitrogen given in May than in March, especially when benodanil was given. In 1974, little rain fell until July and then March nitrogen increased grain yield most whether or not benodanil was given. Top dressing of N (in May) produced a smaller straw yield than seed-bed dressing did in each year.Benodanil spray much decreased brown rust and increased yield of grain, but not of straw; the increase was greater in 1973 than in 1974, when brown rust was less abundant. Benodanil increased yields of all varieties, but that of the rust susceptible variety, Midas, most. Benodanil increased grain yield more, as more N was given to the barley, and more with N given in May than in March. It increased the amount of N removed by the barley grain, but not by the straw.

Description: SummarySeven experiments on winter barley, made during 1979–81, measured the effects of several amounts of nitrogen, applied at different times in spring, on grain yield and nitrogen content. All of these experiments also tested a growth regulator applied in spring, and five of them tested fungicide sprays. In 1980 the experiment at Rothamsted also tested sowing dates, and an eighth experiment, also at Rothamsted in 1980, tested sowing dates and mildew fungicides applied in autumn or spring. In all the experiments treatments were tested in factorial combination.N applied in February was less effective than N applied in March and both were less effective, in terms of grain yield, than N applied in April. Divided dressings were best applied in February and April or March and April; effects on % N in grain followed the same pattern and so, therefore, did the efficiency of uptake of fertilizer N. The growth regulator consistently reduced the length of the straw and diminished lodging; it increased yields in five of the seven experiments in which it was tested.Responses to fungicides were inconsistent from year to year. Sowing in September rather than in October 1979 increased yield in both of the experiments at Rothamsted in 1980, especially of the 6-row variety Hoppel.

Description: SummaryExperiments on winter barley at Rothamsted, testing different sowing dates, were sampled in 1987–89 to measure effects on take-all caused by the fungus Gaeumannomyces graminis var. tritici. The experiments used the same plots in each year, and in 1988 and 1989 the randomization was restricted so that sowing dates were balanced for sowing dates in the previous year.The site had been used to grow barley since at least 1979 (spring barley from 1979 until 1985), and take-all was much more severe than expected. It was usually most severe in the earliest-sown plots and decreased almost linearly as sowing was delayed. There was also evidence that the sowing dates of the preceding crop had a continuing, residual effect. In the two years that the effects of previous sowing dates were tested, there was usually least take-all in plots where the crop followed one sown very early or very late in the previous year, and most where it followed one sown at the end of September or in early October. These effects probably reflect differences in amounts of inoculum and in the rate of development of take-all decline.

Description: SUMMARYResults obtained from the plots of a field experiment may be subject to’ representational errors'; that is, they may not exactly reproduce the results that would be expected if the same treatments were applied to whole fields. In particular, the treatment applied to one particular plot often has the potential to affect the adjacent plots; such an effect may be positive or negative according to circumstances, and its magnitude cannot easily be estimated. This paper shows, on theoretical grounds and by simulations, that nearest-neighbour (NN) analysis cannot be relied upon to mitigate the effects of such inter-plot interference, even when the residual variance of the NN analysis is substantially, and by normal criteria significantly, smaller than that of conventional (randomized block) analysis. In certain circumstances, NN analysis of data affected by interference produces greater average bias in estimated treatment differences than conventional analysis.

Description: SUMMARYDisposal methods for straw from continuous winter wheat were tested on two soil types, a flinty silty clay loam and a sandy loam, over 7 years (1985–91). The methods tested were burnt or chopped straw in full factorial combination with four cultivation methods (tined to 10 cm, tined to 10 cm then to 20 cm; ploughed to 20 cm; tined to 10 cm then ploughed to 20 cm). Measurements were taken to determine the effects on crop establishment and growth, pest and disease incidence, and the consequent effects on yield. Another experiment (1985–91) on the flinty silty clay loam site, investigated the interactions between straw treatments (burnt, baled or chopped in plots that were all shallow cultivated to 10 cm) and five other factors; namely, time of cultivation, insecticides, molluscicides, fungicides and autumn nitrogen. All the straw x cultivation systems allowed satisfactory crops to be established but repeated incorporation of straw using shallow, non-inversion cultivations resulted in very severe grass-weed problems. Early crop growth, as measured by above-ground dry matter production, was frequently decreased by straw residues, but the effect rarely persisted beyond anthesis. Pests were not a problem and their numbers were not greatly affected either by straw or cultivation treatments, apart from yellow cereal fly which, especially on the heavier soil, was decreased by treatments which left much straw debris on the soil surface. Incorporating straw also caused no serious increases in the incidence of diseases. Indeed, averaged over all sites and years, eyespot and sharp eyespot were both slightly but significantly less severe where straw was incorporated than where it was burnt. Eyespot, and even more consistently sharp eyespot, were often more severe after ploughing than after shallow, non-inversion cultivations. Effects on take-all were complex but straw residues had much smaller effects than cultivations. Initially the disease increased most rapidly in the shallow cultivated plots but these also tended to go into the decline phase more quickly so that in the fourth year (fifth cereal crop) take-all was greater in the ploughed than in the shallow cultivated plots. On average, yields did not differ greatly with straw or cultivation systems, although there were clear effects of take-all in those years when the disease was most severe. In the last 2 years, yields were limited by the presence of grass weeds in the plots testing chopped straw incorporated by tining to 10 cm.

Description: SUMMARYExperiments on spring barley at Rothamsted over the years 1974–81 were used to study how sources of powdery mildew inoculum affected responses to fungicide sprays applied at different times. Reinfection of early-sprayed plots was generally faster where they were close to a potent source of inoculum than where they were not. Conversely, effects of sources on yield, although not always significant, occurred mainly in the later-sprayed plots. Sources thus had their most important effects before these later sprays were applied, and probably during the very early stages of the epidemic. This implies that if experiments are to approximate to fields, they should, initially, be part of a large, uniformly-susceptible area of crop.Regression analyses showed that delaying the application of fungicide sprays after the optimum date caused smaller yield losses in plots near minimal sources of inoculum than in plots near more potent sources. There was no conclusive evidence for effects of inoculum sources on the optimum date to apply a fungicide spray but the tendency was for the optimum to become later as inoculum pressure increased.Separating fungicide-treated plots with mildew-resistant barley resulted in larger treatment effects and smaller residual mean squares than where there was no separation. Judged by residual mean squares alone, precision was approximately doubled by separation.

Description: SUMMARYAn experiment at Rothamsted in 1985–89 and another at Whaddon in 1986 studied the effects of incorporating straw on diseases of winter barley. Net blotch (Pyrenophora teres) and leaf blotch (Rhynchosporium secalis) were initially less severe where straw was burnt or incorporated by ploughing than where cultivations only partially buried it. However, by summer both diseases were usually more severe where straw had been burnt than where it had been incorporated. At Whaddon, eyespot (Pseudocercosporella herpotrichoides) tended to be less severe in tine-cultivated plots where straw was incorporated than where it was burnt, but at Rothamsted, where the straw treatments were confounded with cultivations, there was no consistent effect. The disease was usually more severe where straw was incorporated by ploughing than where it was incorporated using other methods. In contrast, the severity of take-all was generally decreased by ploughing. Seedlings usually grew better where straw had been burnt rather than incorporated and grain yields were often larger. However, yields at Rothamsted in 1987 were unusually, and inexplicably, smaller after burning the straw so that the 5-year mean yields showed no significant differences between treatments.

Description: Results from field experiments with mobile pests and air-borne pathogens are subject to bias as a result of inter-plot interference. Serially balanced designs (SBDs) allow interference to be estimated but other designs may be better for decreasing such effects. To investigate this, systematic replicated designs, comparing sprays applied at different times to control powdery mildew of spring barley, were sited in 2 years alongside SBDs testing similar treatments. Yields of grain and assessments of mildew on the leaves were analysed. Results from the balanced designs provided strong evidence of interference in both years but not in a third (when the systematic design was omitted). Estimates of treatment effects from the systematic designs were often, but not consistently, greater than corresponding estimates from the SBDs. A method of analysis from Draper & Guttman (1980) was adapted to produce estimates of the differences between treatments as if applied to all plots of an experiment; this showed larger differences between treatments than the conventional analysis in 1975 and 1976 (when there was appreciable interference), but failed in 1977 when interference was slight. This method fails when applied to the systematic designs; SBDs (which are a subset of all designs in randomized blocks) are probably optimal for this type of analysis. The difficulties of analysing data in the form of percentages or proportions (with consequent non-orthogonality) are discussed.

Description: SUMMARYExperiments with spring barley in 1975–7 tested fungicides applied to control powdery mildew (tridemorph) or brown rust (benodanil) in factorial combination with six amounts of fertilizer N, applied either to the seed bed soon after sowing, as a later top dressing or half at each time.Powdery mildew was the principal leaf disease in all three years. It tended to be increased by increments of N and by applying the N late but much less consistently in the first two years, when soils were very dry for much of the growing period, than in 1977 when amounts of rain were much closer to the long-term mean. Tridemorph significantly increased the number of ears in 1975, mean number of grains per ear in 1976 and 1000-grain weight in all three years; it gave net increases in grain yield of 0·55, 0·68 and 0·41 t/ha, respectively, in 1975–7. Yield response to increasing amounts of applied N was greatly increased where mildew had been controlled by the use of tridemorph, and was better where the N had been divided into two dressings than where it had been applied as a single dose. In 1975 and 1977 the biggest responses to tridemorph were obtained with late N but in 1976 yield was increased most by tridemorph where the N had been applied to the seed bed.Analyses of samples taken in 1977 showed no significant effect of tridemorph sprays on concentrations of either N, P or K in the green crop. By contrast, analyses of grain samples in 1976 and 1977 showed that amounts of N in grain (mg N/grain) were affected by amounts of applied N and by tridemorph, and that there were interactions between these two factors. Concentrations of N in the grain (% D.M.) were also determined by the effects which these factors had on grain size. At small N rates tridemorph mostly increased grain size so that N concentrations were decreased by the fungicide. At large N rates increases in grain size where tridemorph had been applied were accompanied by increases in the N content of the grain (mg N/grain) so that N concentrations were either unaffected (1976) or were increased (1977) by the fungicide. With 90 kg/ha of applied N the fungicide increased the amount of N/ha removed in grain by over 21 % in each year. The apparent recoveries of N in these plots were increased from 66 to 81 % and from 87 to 105%, respectively, in the two years. Tridemorph had no significant effect on concentrations of P or K in the harvested grain but increased average amounts of these nutrients removed in the grain by 17 and 14%, respectively, in 1976 and by 14 and 7% respectively, in 1977.Examination of black and white, infra-red aerial photographs of the experiments showed that, in each year, the brightness of individual plot images was significantly correlated with grain yield.Complex designs without division into blocks are especially vulnerable to positional variation. Alternative methods of adjusting for such positional variation were compared in analyses of grain yields. The potential improvements in precision which might be achieved by the appropriate use of such analyses, and the difficulties of ensuring that unacceptable subjectivity and bias are not thereby introduced into the analyses, are discussed.