Agronomy of the system
After three experimental years it was shown that the system can be managed effectively, i.e. it has been established that it is possible to direct drill wheat into the clover sward and that the crop and legume understorey are compatible. Also the work has shown that a number of successive wheat crops can be drilled into the permanent white clover understorey. No difficulties were encountered in scaling-up the work to field size areas and the cereal plants in the bi-cropped area grew most satisfactorily.

There was no difficulty in establishing the clover understorey during the first year of work in the bi-cropped areas. This understorey has persisted well, having had three successive cereal crops drilled into it and still survives at one site to the present day five years after sowing it. Damage to the cereal crop by slugs and other pests and diseases has been minimal except for stem-base diseases which caused serious losses in grain yield at two sites in the third year, but not whole-crop silage yields.

The dense canopy of the legume has largely kept broad-leaved weed populations suppressed. However, grass weeds, especially Poa spp. were very troublesome at some GB sites and during the period May - August '95 grew at such a rate that they formed dense smothering clumps which resulted in very poor silage and grain cereal yields on bi-cropped areas in 1995. Defoliating the clover/Poa spp. after a period of regrowth following cereal harvest and then treating the area with paraquat at 3kg a.i./ha prior to re-drilling the cereal in autumn 1995 and repeating this procedure in 1996 controlled the grass weed problem simply and effectively. Grain yields from bi-cropped treatments were disappointing because of low numbers of ears/m² and grains/ear. The magnitudes of these two components are determined during the early stages of crop development. Ear population depends on the number of tillers formed before stem elongation while the number of grain sites has been determined by the time the wheat plant has reached terminal spikelet. Lack of nitrogen during this critical phase may have prevented the formation of tillers and grain sites and so reduced the potential yield of these treatments. A small amount of nitrogen (50 kg/ha) supplied to one of the bi-cropped treatments during this period increased the number of ears/m² and the number of grains/ear. In addition seedling densities in the bi-cropped treatments may have imposed a further limitation to ear formation. Although bi-cropped treatments were sown at the same seeding rates as conventional treatments, large inter-row widths led to high seeding densities in bicropped rows. Intense competition for nutrients at this sowing density may have limited the ability of plants to form ears. Late sown winter wheat and all of the spring-sown cereal crops failed to establish. These observations are in line with those made in earlier work. A major conclusion from this work is that the bi-cropping is best and probably only suited to growing winter cereals, and particularly winter wheat.

Three years of experimental work showed that although differences between wheat cultivars exist in their reaction to the presence of the clover, the differences appear to be trivial and of no practical consequence. Choice of wheats for bi-cropping with clover is not restricted to short strawed cultivars. In fact, competition from clover may be more severe in a short than in a tall cultivar during the period of tiller development. This and the shorter straw length itself may reduce the dry matter yield of the whole crop. For grain production a longer straw would make combine harvesting easier. The clover types tested differed less than expected in growth habit. However cv. Rivendel established itself faster than cv. Donna during the first 1-2 years while cv. Donna was growing more vigorously in the last year.

It was concluded that when nitrogen is supplied to crops in the form of slurry, crop uptake may vary considerably depending mainly on the amount of ammonia lost by volatilisation, as also shown by other work on conventional winter cereals. For a wheat/clover bi-crop it would be important to recognise and correct for such nitrogen losses rapidly to avoid yield reductions. Slurry applied at rates of 27-30 tonnes/ha between wheat rows, i.e. on top of the clover strips, did not affect the clover adversely. In all years highest yields were obtained with inorganic chemical fertiliser, probably mainly due to comparatively high volatilisation losses of N from applied slurry. This also indicates that the optimum N application level for the bi-crop may be higher than the amounts actually applied in our experiments. Slurry application whether applied overall or in bands did not affect either the clover or cereal detrimentally in the first two years. However, in year three there was considerable lodging on plots where slurry had been applied.

Economic Evaluation
Using standard prevailing commodity and input prices, a financial comparison was made of the two systems at Long Ashton, data form the field-scale comparison and the relevant treatments from the replicated experiment was used. The figures are averaged over the three years (1994, 95 & 96) with the initial costs of establishing the clover understorey (£153/ha variable costs and £100/ha machinery costs) spread over three years. The comparison does slightly favour the bi-crop, as in 1996 the field-scale bi-crop yields were taken from a resown area within effect a first wheat.

Silage The results indicated that for silage production the gross margin of the bi-crop system is comparable with a conventional wheat arable silage; if the bi-crop system lasts a fourth or fifth season, and thus the clover establishment costs can be spread over a longer period, then the system becomes even more attractive. Furthermore if the clover is established as an undersowing in a spring barley crop, an increased gross margin is feasible. The silage from the two systems is given the same value, but the clover rich (higher protein) bi-crop could be given a higher value.

There is also a significant saving in machinery operating costs for the bi-crop system, as the wheat is established each year by direct drilling unlike the conventional system of ploughing and seedbed cultivations.

Grain Results for the bi-crop system are not so appealing, variable costs (and machinery costs) are greatly reduced, but so is the yield and hence the output and gross margin. In the case of grain production the clover is a competitor and not part of the output product of the system. Similar results were obtained at Oak Park, the relatively low grain yield of the bi-crop meant that the cost of producing a tonne of grain was unfavourably high at £98 compared with £55 in the conventional system. Although bi-crop grain yields are low at about 4-5 tonnes per hectare compared with conventionally grown wheat, they do bear favourable comparison with yields of organically grown wheat, and have the advantage of continuous (for a few years) cropping of wheat.

The production or output figures value silage at £80/t dry matter, grain at approximate milling wheat price in October each year. Straw is valued at £20/g in 1994 and £25/t in 1995/6. Grazing is valued at 15 sheep/ha at 50p per week each for 4 weeks. IACS payments are not included but if grown on eligible land, the cereal value (1994 £194, 1995 £269 and 1996 £267) should be added to all gross margin figures. Variable costs include fertiliser, seed, agrochemicals etc. and are valued at farm cost each year. Machinery operating costs, cover all operations including harvesting and are given as average farmer cots according to Nix (1993-6).

Pests, diseases and beneficial arthropods
Initial comparison of data from the four sites show clearly developing treatment differences in the abundance of beneficial predatory arthropod groups with increasingly greater numbers of most groups being caught in the bi-cropped habitat. Carabid beetles showed this trend least well. This may be explained by the extreme mobility of adult carabids between experimental treatments. There is evidence that less mobile larval instars of the carabid Nebria brevicollis became substantially more abundant in the bi-crop system. Staphylinid beetles and linyphiid spiders showed a more marked increase in abundance in the bi-crop system with, particularly the commonest spider populations, showing much greater incidence in the bi-crop compared with the conventional crop system.

In the first crop season there was no observable difference in the numbers of slugs on the bicropped and conventional areas and yet substantially lower levels of slug damage to the cereal was noted in the bi-cropped system. This suggests that the clover acts as an alternative food source for slugs, so reducing cereal injury. In the second, and particularly third crop cycles, slug abundance markedly increased in the bi-crop compared to the conventional area suggesting that increased food and shelter in the perennial clover base, strongly promotes slug numbers. However, the actual incidence of slug injury symptoms on the bi-crop cereal in the second crop was again lower than in the conventional plot suggesting that a slug preference for clover as a food source was still protecting the cereal. In the third crop cycle, slug injury to the bi-crop cereal was considerably more evident at one IE site. In addition to continued increase in slug numbers at this site in this final season, much of the enhanced slug damage seemed to be associated with clover defoliation in the bi-crop plot following the use of herbicide to control grass weeds prior to drilling and the consequently reduced availability of alternative slug food. Grass weed control by herbicide use may therefore be relatively easy to achieve in the bi-crop system, but if broad spectrum herbicides are used this may entail a considerably enhanced risk of slug injury to the following cereal crop in the temporary absence of clover foliage.

Septoria tritici was the only significant cereal pathogen observed in the bi-crop system, at very low levels of incidence over the winter in the first crop cycle, but rather greater incidence in the bi-crop compared with the conventional crop in the second cycle. Despite this, and the reduced use of fungicides on the bi-crop compared with the conventional crop, there was no evidence of increased disease on the bi-cropped cereal later in the season at the flag leaf stage when disease would threaten the crop. Related work showed this to be due to physical prevention of disease spore dispersal up the cereal plant by the clover canopy. In the third crop cycle, Septoria tritici incidence was very low on both crops at all sites.

The incidence of cereal aphids and of aphid transmitted cereal virus disease was low on both crop types throughout the study. Laboratory studies showed that aphid fecundity and populations growth rates under the conditions if lower plant nitrogen levels in bi-cropped cereals were likely to be substantially lower than in conventional cereal crops resulting in less risk of aphid pest outbreaks. Pest and disease incidence on the clover was also very low, although a number of non-cereal aphid species were recorded on the clover plot in the first crop cycle.

Weeds
The dense, smothering clover understorey and cereal crop canopy gave little opportunity for broad-leaved weeds to become a problem. The few dock and thistle plant that occurred on experimental areas were easily controlled by hand-roguing. However, grass weeds were troublesome, especially initially at one GB site. They were easily controlled by the application of paraquat to which cover is tolerant, immediately prior to drilling the wheat in the autumn.

Soil Biology and Fertility
Soil moisture levels monitored under the two crop systems consistently showed lower levels in the bi-cropped plot throughout the study except for short mid-summer periods of very low soil water content in both plots. This effect may be due to greater rates of transpiration under the clover sward and to physical interception of precipitation and loss by evaporation. Whatever the mechanism, lower moisture availability throughout much of the growing season under the bi-crop may have contributed to the lower yields of cereal grain obtained compared with the conventional system. Competition for soil moisture between the clover and cereal may be as important as competition for nutrients. At two sites with heavy-textured soils, there was clear evidence of soil compaction under direct-drilled bi-crops. The bulk density of fine soil (< 2 mm) at 3-8 cm soil depth was between 0.1 and 0.3 g cm^-3 higher under bi-cropping than under conventional cropping.

The greater soil bulk density and reduced soil porosity observed under the bi-crop compared to the conventional plot was to be expected in the absence of soil cultivations. Data obtained from an experiment with earthworm exclusion areas indicate, however, that increased earthworm activity under the bi-crop system has offset this effect to a marked extent, even within the relatively short 3-year duration of the project. The effects of land management and farming practices on earthworm communities are usually studied for two reasons. First, since earthworms are known to enhance soil structure and fertility and plant productivity, it is of interest to know how abundant they are and how land use practices can be modified to enhance their activity. Second, earthworms represent a major component of the soil fauna and they are often regarded as important and useful bioindicators of the general soil biological condition, soil quality, soil health and other biotic concepts that cannot be measured directly. Evidence from this project suggests overwhelmingly that direct-drilling wheat:clover bi-cropping supports much larger and more diverse earthworm communities than conventional wheat cropping. Mean earthworm population densities when averaged across all four sites and all sampling dates (n=160, electrical extraction method) were 578.6 individuals m^-2 and 138.2 g biomass m^-2 under bicropping and 198.2 individuals m^-2 and 35.8 g biomass m^-2 under conventional cropping. At two sites (Long Asthon, GB and Oak Park, IE) with very low initial population levels, the bicropped field had ten times greater earthworm biomass than the corresponding conventional field at the end of the three year study. Trends in the dynamics of earthworm populations at UC Dublin, IE (handsorting method) were well-established after three complete cropping seasons. Population size under bi-cropping grew steadily, whereas that under conventional management was static. The earthworm population in the bi-cropped field was more than three times larger than that in the conventional field, exceeding 1000 individuals m^-2 and 150 g m^-2 throughout the third cropping cycle. These population levels are unusually high for arable land, and are comparable with those reported from productive permanent pastures and similar habitats.

The dramatic increase in earthworm numbers and biomass under wheat:clover bi-cropping is almost certainly mainly attributable to the increased input of high quality food in the form of clover residues and to the reduction of mechanical disturbance. This conclusion is additionally supported by the results from two complementary population studies in replicated tillage and manuring trials. Under all agronomic conditions tested, the electrical octet earthworm sampling method was as good as, or superior to, formalin expulsion which, for reasons of health and safety legislation is no longer a viable method. Replacing formalin expulsion with the electrical sampling technique facilitated the continuation of earthworm population studies at all sites.

Natural abundance measurements of stable N isotope ratios in the field demonstrated that bicropped wheat had used N derived from clover. Earthworms can be expected to accelerate the release of N from decomposing clover root material and from other clover residues, providing mineral N readily available to the crop. Using natural abundance techniques, it was estimated that the Lumbricus populations in the bi-crop mineralised directly between 1.38 and 1.72 g of N₂- derived clover N m^-2 in 180 active days. However, estimates of total earthworm population sizes under the bi-cropping system were often much larger. For example, a earthworm population of 200 g biomass m^-2 feeding on clover residues could mineralise directly between 69 and 86 kg N₂- derived clover N ha^-l in 180 active days. Results from the greenhouse studies based on ¹⁵N labelling suggest that N made available by earthworm activity in an N deficient system benefits primarily the non-legume, wheat. As earthworm populations in these cropping systems are large, this role in N partitioning may be an important mechanism by which earthworms affect the balance between bi-cropped wheat and clover or any other non-legume/legume crop.

Further work
The work developed the system for whole crop silage very successfully and there is now a need to transfer this technology into commercial farming practice via a series of demonstration farm sites. The work has great potential and also needs to be developed for (i) organic cereal production and (ii) for maize, where the clover understorey would prevent soil erosion. Both of these aspects of further development are in progress in GB. The work also needs to be developed to produce cereal grain crops with reduced inputs, but acceptable yields.

Contacts

Coordinator

• Institute of Grassland and Environmental Research / UNITED KINGDOM

Participant

• Villy JOERGENSEN / Danish Institute of Plant and Soil Science / DENMARK
• David KENDALL / Institute of Arable Crops Research Long Ashton Research Station / UNITED KINGDOM
• James I BURKE / TEAGASC - Agriculture and Food Development Authority / IRELAND
• Gordon PURVIS / University College Dublin IRELAND