Given the increasing demand for wheat which is forecast, cropping of wheat in short rotations will likely remain a common practice. However, in temperate wheat growing regions the soil-borne fungal pathogen Gaeumannomyces tritici becomes a major constraint on productivity. In cultivar rotation field experiments on the Rothamsted Farm (Hertfordshire, UK) we demonstrated a substantial reduction in take-all disease and grain yield increases of up to 2.4 tonnes/ha when a low take-all inoculum building wheat cultivar was grown in the first year of wheat cropping. Phenotyping of 71 modern elite wheat cultivars for the take-all inoculum build-up trait across six diverse trial sites identified a few cultivars which exhibited a consistent lowering of take-all inoculum build-up. However, there was also evidence of a significant interaction effect between trial site and cultivar when a pooled Residual Maximum Likelihood (REML) procedure was conducted. There was no evidence of an unusual rooting phenotype associated with take-all inoculum build-up in two independent field experiments and a sand column experiment. Together our results highlight the complex interactions between wheat genotype, environmental conditions and take-all inoculum build-up. Further work is required to determine the underlying genetic and mechanistic basis of this important phenomenon.

Experiments on the Rothamsted and Woburn Experimental Farms studied the effects on take-all of different break crops and of set-aside/conservation covers that interrupted sequences of winter wheat. There was no evidence for different effects on take-all of the break crops per se but the presence of volunteers, in crops of oilseed rape, increased the amounts of take-all in the following wheat. Severity of take-all was closely related to the numbers of volunteers in the preceding break crops and covers, and was affected by the date of their destruction. Early destruction of set-aside/conservation covers was usually effective in preventing damaging take-all in the following wheat except, sometimes, when populations of volunteers were very large. The experiments were not designed to test the effects of sowing dates but different amounts of take-all in the first wheats after breaks or covers apparently affected the severity of take-all in the following (second) wheats only where the latter were relatively late sown. In earlier-sown second wheats, take-all was consistently severe and unrelated to the severity of the disease in the preceding (first) wheats. Results from two very simple experiments suggested that substituting set-aside/conservation covers for winter wheat, for 1 year only, did not seriously interfere with the development of take-all disease or with the development or maintenance of take-all decline (TAD). With further research, it might be possible for growers wishing to exploit TAD to incorporate set-aside/conservation covers into their cropping strategies, and especially to avoid the worst effects of the disease on grain yield during the early stages of epidemics.

The effect of wheat cultivar on the build‐up of take‐all inoculum during a first wheat crop was measured after harvest using a soil core bioassay in field experiments over five growing seasons (2003–2008). Cultivar differences in individual years were explored by analysis of variance and a cross‐season Residual Maximum Likelihood (REML) variance components analysis was used to compare differences in those cultivars present in all years. Differences between cultivars in the build‐up of inoculum were close to or at significance in two of the five trial years (2004 P < 0·05; 2006 P < 0·07), and current commercially listed cultivars were represented at both extremes of the range. In 2007 and 2008, when environmental conditions were most favourable for inoculum build‐up, differences were not significant (P < 0·3). In 2005 the presence of Phialophora spp. at the trial site restricted the build‐up of take‐all inoculum under all cultivars. The cross season REML variance components analysis detected significant differences (range: 3·4–47·8% roots infected in the soil core bioassay; P < 0·01) between the nine cultivars present in all years (excluding 2005). This is the first evidence of relatively consistent differences between hexaploid wheat cultivars in their interactions with the take‐all fungus, and this could give an indication of those cultivars that could be grown as a first wheat crop, in order to reduce the risk of damaging take‐all in a second wheat crop. This phenomenon has been named the take‐all inoculum build‐up (TAB) trait.

Results from a series of crop sequence and single-year experiments that tested different straw and cultivation treatments in a total of 11 site-season combinations confirmed previous evidence that the development of eyespot disease in cereals can be decreased by debris in the seed bed even if that debris includes eyespot-infected stem bases, which are the principal sources of primary inoculum. Two of the experiments, which followed non-cereal break crops and tested the effects of crop debris applied after ploughing or tining on eyespot in winter wheat that was artificially inoculated with Oculimacula spp., provided convincing evidence that the effects can be attributed to the debris per se, and not to any associated husbandry practices. There were often larger effects on disease in summer than in spring or on severity than on incidence suggesting that the effects of debris cannot be explained solely by effects on inoculum or initial infection, and that debris has a disease-suppressive effect.

Seed treatments containing fluquinconazole, silthiofam or a standard fungicide mixture with no activity against take-all were compared in all combinations of sequences in successive second and third winter wheat crops in five field experiments and second to fourth crops in a sixth experiment. Compared with the standard treatment, silthiofam decreased take-all more effectively than fluquinconazole when crops were sampled at tillering. In samples taken in summer, during grain filling, silthiofam often decreased the incidence of take-all (percentage of plants with root symptoms) more than fluquinconazole, but fluquinconazole more effectively decreased the incidence of severe take-all (percentage of plants with more than 75% of their root systems blackened). It is suggested that these differences are a consequence of more effective control of primary infection of roots by silthiofam and of secondary, root-to-root, infection by fluquinconazole. Silthiofam usually increased yield more than did fluquinconazole, perhaps as a consequence of better early protection during tiller and/or spikelet formation. Treatment with either of the fungicides affected epidemic development in the treated crop and in crops grown subsequently. In particular, decreased take-all had the effect of delaying the year-to-year epidemic, so that nontreatment of a subsequent crop resulted in an upsurge in disease. Treatment with either take-all fungicide of a crop grown after a treated crop was relatively effective if the epidemic in the comparable nontreated crop sequence was continuing to increase. It was, however, detrimental if the disease was approaching its peak in the first treated crop, particularly if a treated (fourth wheat) crop was being compared with a similar crop in a nontreated sequence in which take-all decline had developed. These results provide a basis for recommendations for the use of seed treatment fungicides in sequences of wheat crops.

The effects of tillage (ploughing vs minimum tillage) and application of chopped maize stalks on winter wheat cv. Hereward by Fusarium culmorum and F. graminearum were investigated in two 2‐year experiments in eastern England. Supplementary inoculum of each fungus (five isolates) was applied to the ground to the first wheat crop in each experiment. Infection of shoot bases (spring), stem bases (summer) and harvested grain were determined by isolating the fungi on agar media and by quantitative PCR. Stem bases were infected more frequently by F. culmorum (up to 35% where inoculum was applied) than F. graminearum (up to 22% where inoculum applied; otherwise scarce). Despite mist‐irrigation to encourage ear infection, the incidence of pathogens in grain was usually low, always less than 4% for F. culmorum, but up to 30% for F. graminearum where inoculum was applied to the ground. Lack of rainfall probably limited the dispersal of conidia from the ground to the ears. Ascospores, produced by F. graminearum (Gibberella zeae) but not F. culmorum, are less dependent on rainfall for dispersal. Infection of grains by F. graminearum was usually correlated significantly with stem infection and sometimes, negatively, with grain yield. Both conidia and ascospores were produced on the previous crop's debris, apparently in small amounts, throughout much of the year. Little horizontal (plot to plot) dispersal was evident. Non‐inversion cultivation and maize‐stalk amendments tended to decrease stem‐base disease, presumably through the effects of microbial antagonism or competition. However, the incidence of F. graminearum was increased in stem bases where maize stalks were present and in grain by both the presence of maize stalks and minimum tillage. The results support experimental data and UK survey data that non‐inversion tillage increases the risk of ear infection, especially by F. graminearum, and hence of mycotoxin contamination of grain. Maize presents an additional risk where weather conditions do not limit the proliferation of the pathogens. This is consistent with evidence from surveys that the region in which the wheat crop is grown is the main risk factor for ear infection and mycotoxin accumulation in grain.

Take-all disease (Gaeumannomyces graminis var. tritici) in wheat crops is known to be suppressed by naturally occurring antagonistic fungi, closely related to the pathogen, that infect grasses and cereals. This form of suppression was re-investigated because of the changing importance and role of grass weeds and grass covers in arable farming. Natural populations of the competitive fungus Gaeumannomyces cylindrosporus, allowed to develop under rye-grass, were more effective than artificially introduced populations in suppressing the development of take-all in following wheat crops. To be effective, the antagonist needs to be present before the start of wheat cropping. Introducing G. cylindrosporus, but not G. graminis var. graminis (a potential antagonist that is faster growing), into a previous crop, or just after the previous crop, sometimes suppressed take-all, but the effect was small. It is concluded that, for any future attempts at biocontrol by these fungi, they should be introduced into a preceding crop not susceptible to take-all. Take-all inoculum in the soil should be at a minimum and effective hosts of the take-all pathogen must not be present as weeds or volunteers.

Grass species were grown in plots, as pure stands or mixed with wheat, after a sequence of wheat crops in which take-all (Gaeumannomyces graminis var. tritici) had developed. Annual brome grasses maintained take-all inoculum in the soil as well as wheat (grown as a continuous sequence), and much better than cultivated species with a perennial habit. Take-all developed more in wheat grown after Anisantha sterilis (barren brome) or Bromus secalinus (rye brome), with or without wheat, than in continuous grass-free wheat in the same year, where take-all decline was apparently occurring. It was equally or more severe, however, in wheat grown after Lolium perenne (rye-grass) or Festuca arundinacea (tall fescue), despite these species having left the least inoculum in the soil. It was most severe in plots where these two grasses had been grown as mixtures with wheat. It is postulated that the presence of these grasses inhibited the development of take-all-suppressive microbiota that had developed in the grass-free wheat crops. The effects of the grasses appeared to be temporary, as amounts of take-all in a second subsequent winter wheat test crop were similar after all treatments. These results have important implications for take-all risk in wheat and, perhaps, other cereal crops grown after grass weed-infested cereals or after set-aside or similar 1-year covers containing weeds or sown grasses, especially in combination with cereal volunteers. They also indicate that grasses might be used experimentally in wheat crop sequences for investigating the mechanisms of suppression of, and conduciveness to, take-all.

Field isolates (n = 144) of the wheat take‐all fungus Gaeumannomyces graminis var. tritici (Ggt) were tested for sensitivity to silthiofam, a take‐all‐specific fungicide used as a seed treatment, and identified as A‐ or B‐type by PCR–RFLP analysis of nuclear rDNA. A possible association was identified between polymorphisms in ITS2 of the nuclear rDNA and sensitivity to silthiofam. A Ggt‐specific PCR assay was developed which simultaneously identified isolates of Ggt as A‐ or B‐type, based on the polymorphisms in the nuclear rDNA. A highly significant correlation between Ggt type using the PCR assay and sensitivity to silthiofam was demonstrated in a collection of 358 isolates from three field experiments designed to test the effects of seed‐treatment fungicides on take‐all and Ggt populations in winter wheat. In one experiment the percentages of silthiofam‐sensitive and B‐type isolates were significantly less in populations from plots sown with silthiofam‐treated seed in two consecutive years than in populations from plots sown with nontreated seed. However, silthiofam still provided a significant amount of control of take‐all. The natural occurrence of fungicide‐insensitive isolates, up to about 30% in soils in which the fungicide had never been used, is unusual. The new PCR assay provides a useful tool for studying the population structure of Ggt, and may provide a novel method for assessing the incidence of insensitivity to silthiofam (the target site for which has not yet been identified) in field populations of Ggt.

The susceptibilities of different grass species, including currently important annual weeds of cereal crops, to root infection by Gaeumannomyces cylindrosporus and related weakly or non-pathogenic fungi, and to G. graminis var. tritici (the take-all fungus), were tested in pot experiments. Amounts of infection on wheat grown subsequently were also compared. Infection by the non-take-all Gaeumannomyces spp., arising from artificial inoculation, was variable but characteristic swollen cells were often more numerous in root systems of wheat than of grasses. Take-all, arising from natural soil infestation, in the following wheat was decreased only after inoculation of the grasses with G. graminis var. graminis in one experiment; this effect was not influenced by previous host species. Wheat became more affected than grasses by take-all but bromes (Anisantha sterilis and Bromus secalinus) were more diseased than the other grasses. Overall, annual grasses, including the bromes and black-grass (Alopecurus myosuroides), developed more take-all than perennial grasses. The most affected species allowed most take-all to develop on wheat plants grown subsequently. Where a second sowing of wheat was tested (i.e. a total of three sowings), take-all after the perennial grasses was similar to or more than that after annual grasses or wheat. Implications for the effects of weed grasses on take-all in modern cereal cropping systems, particularly those involving set-aside, are discussed.

A seed treatment containing fluquinconazole as the only active ingredient was tested in sequences of up to six consecutive crops of winter wheat. It was applied or not applied in each year, and was tested in all possible combinations with treatments applied in previous years. Take-all was controlled effectively, and grain yield usually increased, when the disease intensity was moderate or severe in non-treated crops, but control of the most severe take-all did not result in acceptable yields or grain quality. Treatment of a first wheat or second wheat with little take-all did not usually benefit the subsequent crop. Non-treatment of a crop grown after a treated, diseased crop usually resulted in a marked increase in disease, indicating that treatment had delayed progress of the epidemic. Take-all was controlled by treatment of a crop grown after a treated, diseased crop but the amount of control and of increased yield was often less than that in a treated crop grown after a non-treated crop in the same crop sequence. Similar effects of seed treatment were apparent in crops grown on a site with take-all decline. The alternative fungicide, silthiofam, applied as a seed treatment in the later years of some experiments, was usually as effective as fluquinconazole. From these experiments, it is recommended that: a) fluquinconazole seed treatment should be applied to a second or third wheat crop, grown after a first wheat crop that was managed to avoid rapid take-all development (e.g. by avoiding very early sowing); b) a break crop should follow the treated crop; c) the seed treatment should not normally be used in longer sequences of wheat or on take-all decline soil unless it is planned to follow the treated crop with a non-cereal break.

Relationships between take-all intensity and grain yield and quality were determined in field experiments on cereal crops using regression analyses, usually based on single-point disease assessments made during anthesis or grain-filling. Different amounts of take-all were achieved by different methods of applying inoculum artificially (to wheat only) or by using different cropping sequences (in wheat, triticale or barley) or sowing dates (wheat only) in crops with natural inoculum. Regressions of yield or thousand-grain weight on take-all intensity during grain filling were similar to those on accumulated disease (area under the disease progress curve) when these were compared in one of the wheat experiments. Regressions of yield on take-all intensity were more often significant in wheat than in the less susceptible crops, triticale and barley, even when a wide range of disease intensities was present in the latter crops. The regressions usually had most significance when there were plots in the severe disease category. Thousand-grain weight and hectolitre weight usually responded similarly to total grain yield. Decreased yield was often accompanied by a significant increase in the percentage of small grains. When severe take-all was present in wheat, regressions showed that nitrogen uptake was usually impaired. This was sometimes accompanied, however, by increased percentage nitrogen in the grain as a consequence of smaller grain size with decreased endosperm. Significant effects of take-all, both positive and negative, on Hagberg falling number in wheat sometimes occurred. Significant regressions of yield on take-all assessed earlier than usual, ie during booting rather than grain-filling in wheat and triticale and during anthesis/grain-filling rather than ripening in barley, had steeper slopes. This is consistent with observations that severe disease that develops early can be particularly damaging, whilst the crops, especially barley, can later express tolerance by producing additional, healthy roots. The regression parameters, including maximum potential yield (y-axis intercept) and the extrapolated maximum yield loss, also varied according to the different growing conditions, including experimental treatments and other husbandry operations. These differences must be considered when assessing the economic potential of a control measure such as fungicidal seed treatment.

A series of field experiments, each with a minimum of five consecutively grown crops of winter wheat, was used to study responses to fluquinconazole seed treatment applied at different stages in the development of take-all epidemics, including take-all decline. The results form the basis of recommendations for using fluquinconazole seed treatment for managing take-all in sequences of wheat crops.

SummaryIn a field experiment on winter wheat, take‐all on plants and the infectivity of the soil were studied in crop sequences with different combinations of sowing dates. Take‐all was negligible in the first wheat crop, but thereafter the mean disease intensity (measured using a take‐all rating, TAR, with a maximum of 300) was 108, 190, 118 and 251 in the second to fifth successive crops. In each growing season, the disease differed amongst sequences and built up more rapidly and was more intense on plants sown in mid‐September than on plants sown in mid‐October. In late‐sown plots, where volunteers had been present during the mid‐September to mid‐October period, take‐all reached an intensity intermediate between that in early‐sown plots and that in late‐sown plots that had been kept free of volunteers. Volunteers, therefore, partially offset the expected beneficial effect of decreased disease with later sowing. Differences in take‐all amongst sequences were most pronounced in the second wheat crop and early sowing of the previous wheat increased intensity of disease. In the following (third) crop, differences in disease intensity amongst sequences were smaller. Soil infectivity (measured by seedling bioassay after harvest) built up progressively from a low level after the first crop to peak after the third crop. In this build‐up phase, soil infectivity estimates were always numerically greater after harvest of early‐sown treatments than after later‐sown treatments, although never significant at P= 0.05. The greatest difference (P= 0.06) was recorded in October before sowing of the third crop, where the comparison was between soil after two previous early sowings and soil after two previous later sowings and control of volunteers. In the same autumn, presence of green cover (i.e. volunteers) was associated with a smaller loss of soil infectivity between harvest and later sowing than occurred in an absence of green cover. In 2nd–4th crops, where comparisons were available and mean TARs indicated moderate levels of take‐all, sowing later had no yield benefit, despite more take‐all and greater soil infectivity associated with early sowing. Important considerations for the management of crops at risk of take‐all are 1) choosing appropriate sowing dates to minimize take‐all or to encourage take‐all decline and 2) controlling volunteers and weed hosts where crops are sown late to minimise take‐all.

Three experiments on winter wheat, each lasting 5 years and on different soil types, were used to test the effects of incorporating different amounts of straw, mainly to determine the importance of achieving uniform distribution to avoid adverse effects on grain yield. Decreases in crop growth and/or grain yield as a consequence of incorporating straw were detected in the first year. The decreases were much larger in one experiment where straw was imported and applied to soil that had been fallowed for 12 months before sowing the wheat than in the other two where the straw was incorporated following the harvest of a winter wheat crop. In the subsequent 4 years, incorporating up to 20 t straw/ha had no significant effects on grain yield but there were some significant effects on concentrations and uptakes of N, P and K, especially on the heavier textured soils. The effects on crop growth and yield that were detected in the first year on each site are tentatively attributed to decreases in available N representing that which was required to support the decomposition of the incorporated straw. The relative lack of significant effects in subsequent years seems to imply that a significant proportion of this N was remineralized relatively quickly, and thus available to support the decomposition of the straw that was incorporated in the second year and, after further recycling, in the years after that. Eyespot, caused by the fungus Pseudocercosporella herpotrichoides, was decreased by incorporating straw but there were few significant effects on other diseases. The results provide a generally reassuring message for farmers in suggesting that on most, if not all, soils there is little cause for concern about the consequences of incorporating even large amounts of wheat straw before sowing a further crop of winter wheat.

Yield responses to fluquinconazole, applied to the seed of winter wheat in experiments on take-all risk sites in 1996-98, were better related to the yields of untreated plots in the different experiments than to take-all, apparently reflecting, in part, inconsistent effects of the disease on yield. Effects of treatment were not altered by changes in seed rate. In a four-year sequence of winter wheat crops, responses to fluquinconazole in the year of application increased progressively as take-all became more severe. However, applying fluquinconazole to the third wheat, in which take-all was moderate to severe, decreased yield in the fourth wheat compared to plots that followed untreated third wheats, probably because treatment in that year delayed peak disease and the onset of take-all decline. Until more information is available farmers should be cautious about applying fungicides to control take-all to more than two or three consecutive crops of winter wheat, especially if early-sown.

Average percentages of winter wheat plants with severe take-all were decreased by up to half by azoxystrobin applied as foliar sprays in four field experiments. Decreased take-all in three of the experiments was associated with increased grain yield but effects on other diseases may have contributed to these responses. Standard fungicide sprays were ineffective. The effects differed, but not consistently, among different cultivars that were tested in three of the experiments. One, two or three sprays of azoxystrobin or kresoxim-methyl, in autumn, spring or summer, were tested in the fourth experiment. Unlike azoxystrobin, kresoxim-methyl had no consistent effects but a smaller amount was applied. Two or three sprays of azoxystrobin were more effective than a single spray but their timing was unimportant. Such control of a root disease by a foliar-applied fungicide is unusual but may help to explain some of the unexpectedly large yield responses to azoxystrobin that have been reported. This relatively broad-spectrum fungicide may have the potential to contribute to the practical management of take-all but further research is needed to determine how best to exploit its effects consistently.