1 Introduction *

2 Aims of study *

3 N cycle in grass/white clover leys *

4 N fixation and grass and clover dynamics *

5 Soil chemistry of grass-clover swards *

6 Effects of clover and grass on soil nitrogen *

7 Colonisation of different soil microsites by clover and ryegrass (invasion experiment) *

8 Discussion *

Table 1.4.1 Methods of measuring N fixation *

1. Introduction
1. The nitrogen cycle

After water, nitrogen (N) is the nutrient that most commonly limits plant growth (Vitousek and Howarth, 1991). Every time crops, meat, milk, eggs or wool are exported from a farm, the nitrogen contained within them must be replaced, as this nitrogen has come out of the soil of the farm, and if it is not replaced, soil fertility will suffer. For this reason, the supply of nitrogen is vital for successful agriculture.

In natural ecosystems, nitrogen passes through a cycle that involves living organisms, the soil and the atmosphere. Nitrogen gas (N₂) is the main component of the atmosphere. N₂ is largely inert, but it can be converted into more reactive N compounds by a variety of processes known collectively as N fixation. Certain types of bacteria can convert N₂ into ammonium (NH₄⁺) compounds. These N-fixing bacteria can be either free living or in close associations with plants (symbiosis). Of these the symbiotic bacteria present in the roots of leguminous plants are the most important, both in terms of the amounts of N that they fix globally, and in their importance in agriculture (Schlesinger, 1997). In most environments, N fixation by free living N fixing bacteria is low, usually less than

10 kg ha^-1yr^-1 (Lockyer and Cowling, 1977). However, free living N fixing bacteria can be important in some environments, especially where there is an abundant supply of energy rich organic matter, but little nitrogen. Relatively high levels of non-symbiotic nitrogen fixation often occur in decaying logs (Roskoski, 1980). In the broadbalk continuous wheat experiment at Rothamsted, N fixation by free living bacteria has been estimated at 28 kg N ha^-1 year^-1 (Witty, 1979). Some N fixing bacteria live symbiotically within the hindguts of arthropods, particularly those that feed on decaying wood and other low N diets e.g termites. In some environments, N fixation in arthropod guts may be as high as 10-40 kg N ha^-1 year^-1 (Nardi et al. 2002). Symbiotic N fixation will be discussed more fully in section 1.2. N₂ can also be fixed by lightning, and industrially by the Haber process, which is used to manufacture fertiliser. The Haber process was developed during the First World War, originally in order to manufacture explosives. N is artificially fixed in the form of ammonia, by heating N₂ at high pressure in the presence of a catalyst. The Haber process can be summarised by the following equations:

[.1]

[.2]

(Schlesinger, 1997). In 2000, the world used approximately 80 million tonnes of fertiliser N, in 1960 the figure was nearer 11 million tonnes (International Fertilizer Industry Association, 2000). Agricultural legumes currently fix approximately 40 million tonnes of nitrogen. In pre-industrial times, terrestrial N fixation was approximately 90-130 million tonnes per annum. Human impacts have therefore approximately doubled the inputs of N into the terrestrial N cycle (Vitousek et al., 1997). Burning of fossil fuels also causes a release of fixed N. Vehicle exhausts can also be a significant input of nitrogen in industrialised countries (Lee and Dollard, 1994). By these processes N passes from the atmosphere into the soil, water and living organisms. N is returned to the atmosphere from the soil by the actions of denitrifying bacteria, completing the cycle.

The majority of organisms have no ability to fix nitrogen, and so they are dependent on a N cycle within the larger scale N cycle of fixation and denitrification, in which fixed nitrogen is continually recycled. N fixation only accounts for an estimated 12% of the nitrogen used annually by plants. The rest is obtained by recycling of nitrogen already present in plants and the soil. It has been estimated that the mean turnover time of nitrogen in terrestrial ecosystems, from fixation to denitrification, is around 700 years (Schlesinger, 1997). Within the soil, microbes convert NH₄⁺ into nitrate (NO₃) and nitrite (NO₂), a process known as nitrification. Plants take up

NO₃^- and ammonium NH₄⁺ from the soil nutrient solution and convert this into protein. Some of the plant protein is returned to the soil through death and decay of part or all of the plant. Some plants can utilise amino acids and complex organic N compounds as N sources in addition to nitrate and/or ammonium (e.g. Persson and Nasholm, 2001, Lipson and Nasholm, 2001). In the soil, bacteria convert the dead organic material into microbial biomass and soil organic matter, which includes a complex mixture of nitrogen compounds. Over time, soil organic nitrogenprotein is mineralised by bacteria into amino compounds, then ammonium and finally to nitrate (Bolger et al, 2001, Killpack and Bucholz, 1993). The speed at which this occurs depends upon the ratio of carbon to nitrogen in the soil organic matter. From measurements of the turnover of soil organic matter and inputs, it has been estimated that on average N remains in soil organic matter for over 100 years (Schlesinger, 1997).

When animals feed on plants, some of the plant protein N is converted into animal protein, but most is returned to the soil as dung and urine. Henzel and Ross (1973) compared N intakes and excretion of N by farm animals on diets with a range of N contents and concluded that beef cattle excreted 90-96%, sheep excreted 87-95% and dairy cattle excreted 72-87% of the N in their diet. Ultimately, when animals die, the N contained in their bodies decays and eventually becomes available to plants and microbes. Nitrogen from dung and urine is incorporated into the soil organic matter, and is eventually released into solution as ammonium, nitrate and nitrite. These basic principles apply as much to animals within the soil that feed on roots and decaying plant material, as they do to cows and sheep.

Some of the N in soil is lost to the atmosphere, either by volatilisation of ammonia, or denitrification by bacteria, in which nitrate is converted to nitrite, and then into nitric oxide (NO), nitrous oxide (N₂O) and N₂. The latter three compounds can escape to the atmosphere (Erickson et al. 2001). Denitrification takes place under anaerobic conditions, and is commonest in waterlogged soil, although it can take place in anaerobic pockets within most soils (Schlesinger 1997). Volatilisation losses of N are greatest in fertilised soils and during the decomposition of urea excreted by animals (Terman, 1979, Fillery, 2001). Nitrogen is also lost from the soil when it is leached out of reach of the plant root systems, by rainfall. In natural ecosystems, leaching and gaseous losses are usually small (e.g. Erickson et al. 2001). A simplified overview of these processes is shown in Figure 1.1.1



Figure .1 The Nitrogen Cycle

1. Symbiotic Nitrogen fixation

As we have seen, certain bacteria have the ability to fix atmospheric N. Some plants have developed associations or symbioses with these bacteria, allowing the plant to benefit from the fixed N. The bacteria generally benefit from the relationship because the plant provides an environment in which they can thrive. In some cases the association is a loose one, in which N fixing bacteria simply live around the roots of the plant, feeding off root exudates. Examples of this include the N fixing bacteria Azospirillum and Azotobacter, which are found associated with the root systems of tropical grasses. Sugar cane can be grown using these bacteria as an N source (AbdelMonem et al., 2001, Oliveira et al. 2002).

A closer association is between the aquatic fern Azolla and the nitrogen fixing cyanobacteria Anabaena. Here the two species have evolved together to such an extent that the fern has developed hollow leaves in which the cyanobacteria live. Azolla is an important source of N for many rice farming systems in the tropical and subtropical areas of the world (Ferentinos et al. 2002).

The main group of plants capable of forming symbiotic relationships with N-fixing bacteria, are the legumes. The majority of the Leguminosae are capable of fixing nitrogen by forming symbioses with bacteria of the genus rhizobium. Rhizobia infect the root hairs of legume plants, and root nodules form, which provide an anaerobic environment in which the rhizobium can live. The legume provides the bacteria with sugars as an energy source, the product of photosynthesis, and utilises the N fixed by the rhizobium. As with any mutualistic relationship between species, there are costs and benefits to both species. In particular, maintaining the symbiosis has energy costs for the legume, with the effect that it is more energetically expensive for the legume to fix nitrogen than to utilise soil N. This suggests that the ability to fix N will be more advantageous in situations where N is limiting. Certain trees, notably Alder form associations with the actinomycete Frankia, which are similar to the associations between legumes and Rhizobia.

3. Nitrogen in Agriculture
1. Nitrate, ammonium and SON

It is often stated that "When taken up by plants, nitrate and other nutrients are identical in form whether they come from organic matter, soil reserves or applied fertiliser." (e.g. Fertiliser Association of Ireland, 2002). This does not mean that clover has precisely the same effect on soil as N fertiliser. For instance, legume roots can aerate soil (Patriquin et al., 1995) and improve soil structure (Mytton et al., 1993, Miller and Jastrow, 1996). In practice, there are often major differences between legume based and fertiliser based farming systems. In Britain, legume based farming is likely to involve ley-arable rotation, and the use of animal manure (Philipps and Stopes, 1995), neither of which are necessary in fertiliser based farming. The effects of these factors on soil, plants and the environment are discussed in Section 1.2.8. Also, nitrate is not the only source of nitrogen in soil: ammonium and soluble organic nitrogen (SON) compounds can also be used by plants (Lipson and Nasholm, 2001). A popular dictum in organic farming is "feed the soil not the plants" (e.g. Flowerdew, 1998). Nitrate fertiliser is intended to feed crops in the short term, animal and green manures are intended to improve the soil, producing long term benefits to the farm ecosystem as a whole. The use of N fertiliser does not rule out the addition of organic matter to the soil, in the form of green manures or crop residues (e.g. Yadav et al., 2000), and it is possible that organic matter inputs could be increased in this way. However, in recent decades there has been a tendency for N fertiliser to replace other sources of N in agriculture in the developed world. (Uhlin, 1998).

SON is difficult to characterise, but some of it may be amino acids, produced by the breakdown of plant protein. Concentrations of SON in Scottish soils are highest in the presence of plant roots, and removal of vegetation causes this SON to be mineralised into nitrate (Chapman et al. 2001), which strongly suggests that much of this SON is derived from plant material. Plants may be able to access SON after it has been converted to nitrate or ammonium by microbes, and also in some cases by direct uptake. The ability to utilise SON has been observed in plants found in natural ecosystems all over the world (Lipson and Nasholm, 2001), but it has been observed in crop plants as well. Owen and Jones (2001) fed wheat plants with three amino acids, and observed that the wheat plants were only able to utilise 6% of the amino acids, the remainder was absorbed by microbes. Whalen et al. (1999) used ¹⁵N labelled dead earthworms to investigate uptake of organic N by perennial ryegrass (Lolium perenne), and found that the organic N pool was depleted significantly more rapidly in the presence of perennial ryegrass. Jones and Darrah (1993) found that maize plants were capable of using amino acids as an N source, and Matsumoto et al. (2000) found evidence that carrot (Daucus carota) and chingensai (Brassica campestris L.) were capable of utilising soluble organic N compounds from the soil.

The behaviour and role of organic nitrogen compounds in the nitrogen cycle is poorly understood. In conventional farming, N is mostly supplied in the form of nitrate or ammonium compounds (or as urea), and the intention is to feed N to the crop plants directly. In organic and low input agriculture, much of the nitrogen is supplied as organic N compounds, in the form of manure, compost, crop residues etc. Organic farming aims to maintain the soil organic matter content so that the soil can supply crops with the N (and other nutrients) that they need. SON may be an important part of this process, and may also be involved in the transfer of nitrogen from legumes to non-legumes.

Organic agriculture is dependent on a cycle of nitrogen that flows through legumes, non-legumes, animals, microbes, air and soil. It is based on a "systematic connexion or co-ordination of parts in one whole." (Scofield, 1986). For this reason, an understanding of the processes of the nitrogen cycle within the farm is extremely important

2. Legumes as a source of N for agriculture

For the reasons explained in section 1.2.1, comparing the use of legumes and fertiliser N as N sources is complicated because legumes may affect soil in other ways besides simply adding N, and this N may be in different forms to the N applied as fertiliser. Another complication is that the use of legumes in agriculture may require or encourage radically different methods of farming from those associated with fertiliser use.

A wide variety of methods can be adopted to utilise the N fixed by legumes. Legumes in agriculture perform two main functions:

1) Providing protein for human or animal consumption

and/or

2) Providing fixed nitrogen for the benefit of other crops.

When a grain legume such as soybean, is grown and harvested at full seed ripeness, the majority of the fixed N is present in the harvested crop, and soil N may even be depleted by the crop (Toomsan et al., 1995, Haynes et al., 1993). This is not the case with all grain legumes, faba beans (Vicia faba) have been observed to fix 3 times as much N as is removed in harvested grain (Rochester et al., 1998). In these cases, the primary purpose of growing a grain legume is to provide protein (obtained as far as possible from biological N fixation) for human or animal consumption, but residues of the crop may provide fixed N for the successive crop. If the grain legume is fed to animals (or humans) and the animal (or human) faeces are returned to the soil, then much of the fixed N in the grain becomes available to other crops. Grain legumes can be grown in alternate rows with a non-N-fixing species (intercropping), or sown in a mixture with a non-N fixing grain (mixed cropping). This often has the effect of reducing the availability of soil N to the legume, and thus forcing it to fix a higher proportion of its N (see section 1.6). In addition, the non-fixing species may obtain extra N by N transfer (see section 1.2.3). Intercropping and mixed cropping may result in increased yields of one or both species, or a higher land equivalent ratio. The land equivalent ratio (LER) is the area of monoculture of two species required to produce yields obtained in the mixed crop (Helenius and Jokinen, 1994, Fujita, et al., 1992).

Legumes can be grown as green manure crops, and ploughed into the soil in order to improve its nitrogen status prior to sowing of a non-fixing crop (Schmidt et al. 1999). Forage legumes such as clover can be undersown with a taller crop. This provides similar benefits to mixed cropping and intercropping, and can also benefit subsequent crops, smother weeds and provide a habitat for the benevolent, predatory carabid beetles (Armstrong and McKinlay, 1997, Brandsaeter et al., 1998). Perennial legumes, especially leguminous trees, can be cut for composting or mulching (Sanginga et al. 1995), although this is seldom practised in the UK. Cereals can be drilled into an established clover sward (Schmidt et al. 2001, White and Scott, 1991).

In a ley-arable crop rotation, grass and a forage legume such as clover are grown for several years, grazed by animals and/or cut for silage or hay, before being ploughed up so that arable crops can be sown. The last arable crop to be grown before the ley phase can often be undersown with clover. The grass in the ley phase obtains additional N from the presence of the clover; the grazing animals obtain a high protein diet. Manure from the grazing animals transfers fixed N to grass and can be spread on arable crops, which also benefit from the accumulated fixed N that is released when the ley is ploughed (Philipps et al., 1996; Høgh-Jensen, 1996). Ley-arable crop rotations were the main type of agriculture in Europe before the invention of the Haber-Bosch process, for example the Norfolk four course rotation (Lampkin, 1990). This thesis focuses on white clover in ley-arable rotations.

 

3. Nitrogen Transfer from legumes to non-legumes

Atmospheric nitrogen is transferred from legumes to non-legumes either by the so-called "below ground pathway", through root exudation and decay of plant tissue (including foliage), or the "above ground pathway", through dung and urine produced by grazing animals. The below-ground pathway is poorly understood. Nitrogen may be released from the legumes by the senescence and decay of leaves, roots etc. Transport of nutrients between plants via mycorrhizal networks has been suggested (Haystead et al., 1988), although the general view is that transfer is a complicated pathway involving a number of organic compounds. Root herbivory by larvae of the Sitona weevil, and damage to clover root systems may enhance nitrogen transfer (Murray and Hatch, 1996, Hatch and Murray, 1994). Defoliation can accelerate root nodule turnover, and this may be one route by which nitrogen is transferred (Ryle et al. 1995). Dubach and Russelle (1994) have shown that when roots and nodules of alfalfa (Medicago sativa) and Birdsfoot trefoil (Lotus corniculatus L.) are shed, they still contain significant quantities of nitrogen (most other plants efficiently remobilise nutrients from senescing roots and leaves). Estimates of rates of nodule turnover are currently unavailable.

4. Synthetic N fertiliser

In the post war period in the developed world, nitrogen fertiliser has largely replaced nitrogen fixing plants as sources of N in agriculture (Lanyon, 1995). The main exception to this is the soybean, which is widely grown for animal feed, without the use of N fertiliser. Cereals are mostly grown with N fertiliser, high quality pasture is largely maintained by N fertiliser and even some grain legumes are grown using N fertiliser (Redden and Herridge, 1999). The transition to N fertiliser has had numerous environmental and economic effects:

Nitrogen fertiliser has enabled different regions to specialise in crop or animal production (Granstedt, 1991). The nitrogen exported from arable farms, as animal feed no longer has to be returned as manure. In some cases, this has created high concentrations of manure on livestock farms, which must be disposed of (Lanyon, 1995; Granstedt, 1991).

5. Organic Agriculture

Organic agriculture has been defined a number of different ways. The UN Food and Agriculture Organisation (FAO) define organic agriculture as follows:

"Organic agriculture is a holistic production management system which promotes and enhances agro-ecosystem health, including biodiversity, biological cycles, and soil biological activity. It emphasises the use of management practices in preference to the use of off-farm inputs, taking into account that regional conditions require locally adapted systems. This is accomplished by using, where possible, agronomic, biological, and mechanical methods, as opposed to using synthetic materials, to fulfil any specific function within the system." (FAO/WHO Codex Alimentarius Commission, 1999). Organic agriculture does not permit the use of synthetic N fertiliser, instead relying on biological N fixation as its principal source of N. Organic agriculture also prohibits the use of synthetic pesticides and growth regulators. In Britain, most organic farms consist of ley-arable rotations (Philipps and Stopes, 1995). Biodynamic agriculture is a form of organic agriculture based on the ideas of Rudolph Steiner. Biodynamic agriculture requires adherence to a stricter set of guidelines than ordinary organic farming, and additional practices such as the use of special plant based preparations.

6. Energy efficiency of fertiliser N and legume-based systems

Energy efficiency can be defined simply as the amount of energy required to produce a unit of calorific energy in food. However, for practical reasons it is important to consider whether the energy inputs come from a renewable or non-renewable resource. Sunlight, although variable is not going to run out, and use of human labour creates jobs and brings other benefits, even though it may be financially expensive. The definition of energy efficiency should therefore take into account the amount of non-renewable energy required to produce a unit of food or other produce (Jones, 2002). For the purposes of this study, energy efficiency will be defined as the amount of non-renewable energy required per unit of produce.

In the past, the energy used in agriculture derived mostly from human or animal muscle power, which in turn was derived from the calorific energy in the human or animal food (Pimentel, 1979). All of this energy ultimately derived from solar energy, converted into calorific energy by photosynthesis. Modern agriculture in the developed world often replaces human and animal labour with machines such as tractors, which obtain their energy from fossil fuels (although in future, biofuels, such as biodiesel could become more important). Tilling one hectare of soil by human labour requires 400 man-hours and 8.4 ´ 10⁸ joules ha^-1 of energy. A pair of oxen reduces the man-hours to 65, but increase the total energy input to 1.2 ´ 10⁹ joules ha^-1 because of the energy that must be provided to the animals as fodder. A 50 hp tractor reduces the man-hours to 4 but increases the total energy input to 2.3 ´ 10⁹ Joules ha^-1, because of the energy that must be supplied in fuel and the manufacture of the machinery (Pimentel, 1979).

Manufacture of fertiliser consumes fossil fuels. Natural gas is used as the source of hydrogen, and fossil fuels provide much of the energy required for the process. In 1975, corn production in the USA required, on average, 128 kg of N fertiliser per hectare, the manufacture and transport of which consumed 7.9 ´ 10⁹ joules ha^-1. Total energy inputs per ha (labour, machinery, fuel, N, phosphorus (P), potassium (K), lime, seeds, irrigation, pesticides, transport, electricity and processing), consumed a total of 2.7 ´ 10¹⁰ J ha^-1, of which labour comprised 2.3 ´ 10⁷ J ha^-1. The energy efficiency measured as energy output in produce/energy input was 2.93. In 1945, the equivalent figure was 3.7, primarily because increases in energy input have not been matched by increases in productivity (Pimentel, 1979). Fertiliser use accounts for approximately 1/3 of the energy use in US agriculture (Pimentel, 1983). Wittwer (1978) commented that "The current and projected natural gas dependency of chemically fixed nitrogen fertiliser remains as one of the most flagrant violations of good economics, use of a non-renewable resource. It is inconceivable for us to continue to go this route". However, production of ammonia based fertilisers has become more energy efficient over recent years, owing to improvements in the catalysts used. In 1967 4.94 ´ 10⁷ J were required to produce a kg of N fertiliser in the form of calcium ammonium nitrate. In 1998 the equivalent figure was 3.53 ´ 10⁷ J per kg of N fertiliser (Hulsbergen et al. 2002).

These high inputs of fossil fuel derived energy can potentially be reduced by the use of biologically fixed N. Nitrogen fixing plants obtain all of the energy required for N fixation from sunlight. The process does not consume any non-renewable energy sources (Ryle et al., 1979). Laboratory studies have shown that the nitrogenase enzyme (used by rhizobium to fix nitrogen) requires only 1.2 ´ 10⁷ J per kg of N fixed, about 35% of the energy required by the Haber-Bosch process (Postgate, 1979). A survey by Pimentel et al. (1983) estimated energy inputs and outputs (calorific energy in food) in conventional and organically grown corn, spring wheat, potatoes and apples, receiving equal amounts of N as fertilisers, FYM, sewage sludge or green manure. The energy inputs in the form of machinery, fuel, electricity, fertilisers, seeds, pesticides etc. for corn were 2.6 ´ 10¹⁰ J ha^-1 in the conventional system. For conventional wheat inputs were 1.1 ´ 10¹⁰ J ha^-1, for conventional potatoes 6.6 ´ 10¹⁰ J ha^-1 and for conventional apples 1.1 ´ 10¹¹ J ha^-1. The inputs for maize in the organic systems ranged from 1.5 ´ 10¹⁰ J ha^-1 to 1.8 ´ 10¹⁰ J ha^-1; organic wheat 7.2 - 7.8 ´ 10⁹ J ha^-1; organic potatoes 3.5 - 3.9 ´ 10¹⁰ J ha^-1 and for organic apples 8.4 - 8.8 ´ 10¹⁰ J ha^-1. The organic systems therefore used less fossil fuel energy than the conventional systems. The ratio of energy output to energy input in the conventional corn system was 4.47; in the conventional wheat 2.38; conventional potatoes 1.28 and conventional apples 0.89. In the organic systems the equivalent figures were 5.75-7.6, for organic corn, 3.22-3.49 for organic wheat, 1.07-1.2 for organic potatoes and 0.06 for organic apples. The energy efficiency of organic cereal production was therefore greater than that of conventional cereals. The reduced energy efficiency of organic potato production was due to the need for mechanical weed control and the reduced yields. The reasons for the extremely low energy efficiency in organic apple production will be discussed later. The organic systems required between 4 and 64% more human labour. Yields of conventional corn were estimated at 8 t ha^-1, conventional wheat, 1.9 t ha^-1, conventional potatoes 33 t ha^-1 and conventional apples 41.5 t ha^-1. Yields of organic corn were estimated at 7.9 t ha^-1, organic wheat, 1.8 t ha^-1, organic potatoes 16.5 t ha^-1 and organic apples 2.1 t ha^-1. The exceptionally low estimate for organic apple yield is based on the assumption that without pesticides 95% of the apples would be unsuitable for sale. These high losses due to cosmetic damage are also responsible for the low energy efficiency of the organic system (Pimentel, 1983). In contrast Reganold et al. (2001) found that the energy output/input ratio for conventionally grown golden delicious apples was 1.11 and in an organic system 1.18. This latter study ignored the effect of russeting (roughening of the skin) on marketability of apples, on the grounds that, while organically grown apples had a higher level of russeting, this did not make them unacceptable to consumers. In some varieties of apples and pears, russetting is a desired trait (Reganold et al. 2001).

The studies described above only considered the yields of individual crops. In an organic system, arable crops will usually be grown as a rotation, and the efficiency of the rotation as a whole must be considered. Mader et al. (2002) found in 21-year trial of conventional organic and biodynamic ley-arable rotations in Switzerland, that yields were reduced by 20% in the organic systems. However, total energy inputs were 2.1-2.4 ´ 10¹⁰ J ha^-1 year^-1 in the conventional systems but only 1.3 ´ 10¹⁰ J ha^­1 year^-1 in the organic systems. It is important to note that this was a comparison between organic and conventional ley arable rotations, not between fertiliser and ley arable systems. In all systems there was a need for energy to be expended ploughing leys. Improvements in soil structure caused by clover may reduce this energy requirement (Mytton et al. 1993).

In the above cases, the reason for the generally higher energy efficiency in organic systems was the energy used in the manufacture and distribution of fossil fuels, but the organic systems required additional energy to be used in mechanical weed control etc. This, combined with the lower yields reduced the energy efficiency of the organic systems. It is reasonable to examine the potential for making conventional agriculture more energy efficient, without adopting all of the rules of organic farming. Hulsbergen et al. (2002) examined the energy efficiency of various rates of N fertiliser on a rotation of 5 arable crops: potatoes, winter wheat, winter barley, sugar beet and spring barley in central Germany between 1968 and 2000. The study considered inputs of fossil fuel energy, but ignored human labour. In order to achieve the optimum energy output/input ratio for potatoes, the optimum N rate was 54-101 kg ha^-1, which achieved an output/input ratio of 4.7-5.2. For winter wheat, the optimum N rate was 46-54 kg N ha^-1 producing an output/input ratio of 10.5-13. For winter barley, the optimum N rate was 85-102 kg N ha^-1, producing an output/input ratio of 8.7-10.5. For sugar beet, the maximum energy output/input ratio was achieved with 0 kg N ha^-1, producing an output/input ratio of 15-15.2. For spring barley the optimum N rate was 60 kg N ha^-1, producing an output/input ratio of 7-7.2 kg N ha^-1. When by-products such as straw are included, the output/input ratios are even higher (Hulsbergen et al. 2002). The energy efficiency values of cereals and potatoes obtained in the aforementioned study are noticeably higher than the values obtained by Pimentel et al. (1983) for both organic and conventional systems. This may be partly due to the increased efficiency of N fertiliser production (estimated at 50 MJ kg^-1 by Pimentel et al. 1983, and 35.3 MJ kg^-1 in the quoted data from Hulsbergen et al. 2002). The lower energy efficiency in the study by Pimentel et al. (1983) may also mean that maximum energy efficiency may not be the same thing as maximum profitability (see section 1.2.7). Currently the average rates of N application to the principal arable crops in the UK are 189 kg N ha^-1 for winter wheat, 114 kg N ha^-1 for spring barley, 149 kg N ha^-1 for winter barley, 175 kg N ha^-1 for maincrop potatoes, 196 kg N ha^-1 for oil seed rape and 106 kg N ha^-1 for sugar beet (Defra, 2001). The yields of these crops per hectare in 2000 were 7.26 tonnes ha^-1 for wheat, 5.18 tonnes ha^-1 for spring barley, 6.87 tonnes ha^-1 for winter barley, 54 tonnes ha^-1 for sugar beet and for potatoes 41.11 tonnes ha^-1 (DARD, 2001; National Statistics online, 2003). The equations for N fertiliser response calculated by Hulsbergen et al. (2002), underestimated British potato yields by 30.9%, wheat yields by 31%, sugar beet by 2.6% and spring barley by 18%. Winter barley yields were overestimated by 9.6% (equations [1.3.1] – [1.3.5], below). Assuming that energy expenditure on machinery use etc. is the same on British and German farms, the energy output/input ratios for the British arable crops are as follows: potatoes 6.7, winter wheat 7.9, winter barley 8.4, sugar beet 13.2 and spring barley 7. This assumes 20% dry matter in potatoes (Dobozi and Lehoczky, 2002) and 25% dry matter in sugar beet (British Sugar, 2003). These figures suggest that for some arable crops, such as winter wheat, British farmers are currently using higher levels of N fertiliser than the optimum for maximum energy efficiency.

Would the energy efficiency of the arable crops studied by Hulsbergen et al. (2002) be improved if N was supplied by a grass-legume leys rather than fertiliser? To calculate this, it is necessary to know how much N is supplied to subsequent crops by a ploughed ley, and the amount of energy expended in ploughing the ley. Nevens and Reheul (2002) found that a ploughed 3-year-old ley in North West Europe can add 231 kg N ha^-1 to successive crops over 3 years (150, 52 and 29 kg N ha^-1 in the first, second and third years respectively after ploughing of the ley). Adams and Jan (1999) observed that a subsequent crop of ryegrass was able to utilise 30-100 kg N ha^-1 released by the ploughing of a grass-clover ley. The ley phase of a ley-arable rotation produces, meat, milk, wool etc. as well as adding N to the soil, and the energy inputs and production should be calculated separately from the inputs and outputs of the arable phase. The energy required to plough a ley, and prepare a seed-bed from it, prior to planting of an arable crop can be assumed to be equivalent to the "fertiliser energy input", and will be referred to as "ley energy input". Mouldboard ploughing to a depth of 300 mm requires 27.6 L of diesel ha^-1, equivalent to 1.1 ´ 10⁹ J ha^-1 (Sijtsma et al. 1998, Hulsbergen, et al. 2002). Ploughing to 200 mm as favoured by some organic farmers requires 20 L of diesel ha^-1 or 7.9 ´ 10⁸ J ha^-1, because shallow ploughing requires less traction than deep ploughing (Kouwenhoeven et al. 2002, Hulsbergen, et al. 2002). Subsequent seedbed preparation requires between 1.6 ´ 10⁶ J ha^-1 and 4.9 ´ 10⁵ J ha^-1, depending on the machinery used (Chamen et al. 1998), giving a total of between 1.3 ´ 10⁹ J ha^-1 and 2.8 ´ 10⁹ J ha^-1 depending on tillage methods. Assuming that ploughed leys provide between 29 and 150 kg N ha^-1, we can then use the N input/energy yield curves of Hulsbergen et al. (2002), to calculate a figure for energy output/ energy input. This will give an estimate of the energy efficiency if leys rather than fertiliser had been used. The N input/energy yield curves (averaging data from 1989-93 and 1994-98 rotations) are:

Potatoes:

[.3]

Winter wheat:

[.4]

Winter barley:

[.5]

Sugar beet:

[.6]

Spring barley:

[.7]

Where E is the energy output in the crop and N is the N input either from fertiliser or from ploughed leys

(Hulsbergen et al., 2002, average of 1989-93 and 1994-98 rotations)

The figures for N input/ley energy input for crops grown after ploughing of a ley are then 4.1-5.9 for potatoes, 9.8-12.9 for winter wheat, 5.7-12.4 for winter barley, 12.9-15.1 for sugar beet and 4.4-7.7 for spring barley, similar to the maximum energy efficiencies calculated by Hulsbergen et al. (2002). This suggests that the production of arable crops in a ley arable rotation could potentially be as energy efficient or in some cases more energy efficient than production from all-arable systems.

Energy efficiency of arable crops in a ley-arable rotation depends partly on the amount of N transferred from the ploughed ley to the subsequent arable crop, and the amount of energy expended in ploughing the ley. Rather than looking at the arable stage of the rotation in isolation, it is important to also consider the energy efficiency of the ley phase of the rotation. Mader et al. (2002) calculated that a three year biodynamic clover ley requires 15.2 ´ 10⁹ J of energy input over its lifetime. An organic three year ley requires 18.4 ´ 10⁹ J, and a conventional ley requires 22 ´ 10⁹ J, or 39 ´ 10⁹ J, if it receives only mineral N (no manure). Most of the additional energy in the conventional leys was due to fertiliser use (FIBL, 2002). In Britain in 2001, the average rate of N application to grasslands was 130 kg ha^-1 for grazed land, 165 kg ha^-1 yr^-1 for silage plots and 85 kg ha^-1 yr^-1 for hay plots (Defra, 2001). This amounts to 4.5 ´ 10⁹ J ha^-1 yr^-1, 5.8 ´ 10⁹ J ha^-1 yr^-1 and 3 ´ 10⁹ J ha^-1 yr^-1 of fossil fuel derived energy for grazed, silage and hay producing land respectively. Is this input of non-renewable energy justified by the yields?

Attempts have been made to calculate the benefit to grass from clover, in terms of the amount of nitrogen fertiliser required producing the same yield of grass in the absence of clover. This is known as the fertiliser N equivalent. Estimates of fertiliser N equivalent range from 124-275 kg N ha^-1 (Royal Society, 1983). Another study found that grass/legume swards are as effective as grass monocultures receiving 150-200 kg N ha^-1 year^-1, in terms of lamb production, and superior to the monocultures in the post-weaning period. The grass/legume sward could support fewer animals, but the improved performance of the individual animals compensated for this (Frame et al., 1998). The results of these studies strongly suggest that biologically fixed N could replace much of the fertiliser N currently used on grassland in the UK, reducing the fossil fuel energy inputs significantly without affecting production.

Ledgard et al. (1999) compared N inputs and outputs from dairy farmlets receiving 400, 200 and 0 kg N ha^-1 in New Zealand over 3 years. N fixation in the 0N farmlet was 99-231 kg N ha^-1, whereas in the 200 N farmlet it ranged from 75-149 kg N ha^-1, and in the 400 N farmlet N fixation ranged from 15-44 kg N ha^-1. It can be seen from this that inputs of biologically fixed N are reduced in proportion to inputs of mineral N. This subject will be covered in more detail in section 1.6. Dairy farms in the Netherlands typically use 331 kg N ha^-1 of fertiliser, 181 kg N ha^-1 in purchased feed and virtually no biologically fixed N (Ledgard et al., 1999). The ratio N output/N input is 0.14. Dairy farms in New Zealand typically use 40 kg N ha^-1 of fertiliser, 4 kg N ha^-1 in purchased feed, and 140 kg N ha^-1 of biologically fixed N. The ratio of N output/N input is around 0.3. Milk production per hectare in the New Zealand system is about 70% of that in the Dutch system (Ledgard et al., 1999).

The examples given above show that simple comparisons of herbage yield between conventional and organic/legume-based grassland are not sufficient for a comparison of energy efficiency or economic efficiency. Grassland containing clover may offer greater nutritional benefits to grazing animals than a pure grass sward, even if the total herbage yields are lower. Clover also has nutritional benefits for grazing animals, having more protein than grass alone, is quicker to digest (Beever and Thorp, 1996), and remains digestible for longer than L. perenne with increasing maturity (Thomson, 1984). Cows and ewes fed on clover produce more milk, and their milk contains more total protein, more casein and has greater coagulum strength. This makes the milk more suitable for cheesemaking, and is probably the reason for the rapid growth of lambs on clover pastures (Thomson, 1984, Newton et al., 1983).

7. Economic benefits of legume-based farming

As we have seen from the examples in section 1.2.6, legumes can be an energy efficient way to add fixed nitrogen to agricultural systems, especially in the case of grasslands. However, in Britain and in much of the developed world, the use of N fertiliser is more popular as a source of N for agriculture. This section will discuss the reasons for this.

Nitrogen fertiliser is a simple and convenient way of adding nitrogen to the soil. Yields obtained with nitrogen fertiliser-based systems are generally larger than those obtained from other systems (Loomis and Connor, 1992). The amount of nitrogen that can be added to soil this way is only limited by the cost of the fertiliser, and by environmental regulations. Potential maximum crop production can be calculated from experiments in which the roots of the plant in question are provided with optimal growing conditions as regards, water, minerals and oxygen, so that yield becomes dependent on incoming light energy. Monocultures of grass treated with nitrogen fertiliser have a theoretical maximum yield of 30 t DM ha^-1 year^-1 in the UK (Alberda, 1971; Cooper and Breese, 1971). Some fertilised grasslands in Northern Ireland are currently producing 25 t ha^-1 (DARD, 2001). In practice, yields are typically in the region of 13-15 t DM ha^-1 year^-1 in temperate climates (Robson, 1982, Loomis and Connor, 1992, Cooper and Breese, 1971). As long as fertiliser remains relatively cheap, relative to the prices of agricultural products, fuel and human labour, as well as the levels of agricultural subsidies, it makes good economic sense for farmers in developed countries to use it. However, hidden costs are associated with pollution resulting from fertiliser use, which must be paid for by society at large. In the UK in 1996, removing nitrate from drinking water cost an estimated £16m (Pretty et al. 2000). Water companies, such as the German Stadtwerke München, now pay farmers to convert to organic farming to reduce pollution of rivers, although the even higher cost of removing pesticides from drinking water (£120m per annum in the UK, Pretty et al. 2000), probably influences these decisions as well.

The use of grass-white clover pastures in New Zealand has shown their economic viability. The production from these pastures is among the most economical in the world (Frame et al., 1998). Other factors also contribute to the profitability of New Zealand grasslands: New Zealand has a very favourable climate that allows white clover to grow almost all year round, large farms. Production from grassland in New Zealand is dependent upon phosphate fertiliser (MAF, 1974). It has been estimated that a total switch to legume-based systems would benefit UK livestock producers by approximately £300m a year (Doyle and Bevan, 1996). The main reasons why legume-based systems have not been widely adopted are technical, rather than economic (Doyle and Bevan, 1996).

One disadvantage of white clover is that its growth is slow, during the spring, so yield benefits are not apparent until the summer months (Sprent and Mannetje, 1996). However, observations at Haughley Research Station showed that in most years organic pastures began to grow earlier than conventional ones (Balfour, 1976). Poor spring growth is probably due to stolon death over winter, and resulting patchiness of clover distribution (Rhodes and Ortega, 1996), as well as the lower temperatures (see section 1.7.5). Addition of nitrogen fertiliser can inhibit the growth and nitrogen fixation of clover, and cause it to suffer from competition with grass (e.g. Høgh-Jensen and Schjoerring, 1994). Another disadvantage is that clover seeds are small, making them hard to establish. Clover seed must be sown no deeper than 13 mm in a fine seed bed, with soil pH no less than 5.5 (Frame and Newbould, 1986) and they are not persistent. Growth can fluctuate dramatically from year to year, and the nitrogen yield is lower than that obtained with nitrogen fertiliser. The reduced N yield makes clover less attractive from an economic point of view (Sprent and Mannetje, 1996; Stewart and Haycock, 1984). Clover can cause bloat (tympanites) in livestock (D’Mello and MacDonald, 1996), and red clover can also cause fertility problems in livestock (Wong, 1973). Plant breeding programs are currently attempting to rectify these problems (Rhodes and Ortega, 1996). In low input farming a strong economic case can be made for the use of legumes. In organic farming they are essential.

8. Environmental benefits of legume-based farming

While inputs of nitrogen have increased since the introduction of nitrogen fertiliser, efficiency of nitrogen use has declined, with consequent pollution of groundwater and the atmosphere. In the post war period, annual artificial N use in Sweden increased from <15 kg ha^-1 to 79 kg ha^-1, but output did not increase accordingly, in 1980, four times as much N was applied to Swedish farms as was recovered in produce (Granstedt, 1991). In the USA, the separation of animal and crop production, and the increase in animal production that occurred as a result of the use of N fertiliser, caused the efficiency of N use to decline. A high percentage of the applied N was lost to groundwater and the atmosphere (Lanyon, 1995). Nitrogen is lost to the atmosphere as NH₃ and N₂O (a greenhouse gas), and losses of N by this rate can be greater in fertiliser based systems than in legume based systems (Jensen and Hauggaard-Nielsen, 2003). When N is also lost to groundwater, it can cause eutrophication of natural habitats (Howarth, 1998) and pollution of drinking water. The high concentration of dung and urine N on animal production farms increases N losses (Granstedt, 1991).

Organic farming is widely believed to cause less N pollution of groundwater than conventional agriculture, but this belief is not always supported by the evidence. A number of studies have shown that legume-based pastures lose less nitrogen to groundwater than nitrogen fertilised pastures producing similar yields (Tyson et al. 1996; Drinkwater et al., 1998; Ruz-Jerez et al., 1995). Goulding (2000) reviewed data on leaching from conventional and organic leys and found that conventional grass-only first year leys leached around 1.5 times as much N as first year organic grass-clover leys, and second year conventional leys leached around 3 times as much N as second year organic leys. However, leaching from first year organic arable crops was around twice as high as leaching from first year conventional crops, owing to the greater N losses resulting from ploughing of grass-legume leys. N losses from second year conventional arable crops were around 1.5 times higher than N losses from second year organic arable crops. In the third year of arable crops leaching losses were around 1.5 times higher in the conventional system. Over the whole rotation, leaching losses of N were only slightly lower from the organic system.

Mixed farming systems also use nitrogen more efficiently than solely arable farms or pastoral farms (Granstedt, 1991). There is however, a risk of nitrate leaching, following the ploughing of pastures (Scholefield and Smith, 1996). Philipps and Stopes (1995) note that while leaching from this stage of the rotation is high, the lower losses during the rest of a typical organic rotation compensate for this, so the average nitrogen loss is 10.3-20.8 kg N ha yr^-1. Kristensen et al. (1995b) recommend that following the ploughing of a grass clover ley, the best crops to grow are those with a high N demand and/or a long growing season, such as sugar beet, brassicas or cereals. Leaching losses can be minimised if ploughing is done in spring, or late in autumn. An alternative is to plough early in the autumn and plant a catch crop (Kristensen et al., 1995b). Eltun (1995) compared N leaching and surface run off from experimental organic and conventional ley-arable rotations in Norway. Conventional arable fields lost 41.32 kg N ha^-1 as opposed to 13.6 kg N ha^-1 from organic arable land. Leaching and run off of N from conventional leys totalled 22.82 kg N ha^-1 and from organic leys 14.78 kg N ha^-1. The fact that in this comparison, organic leys were ploughed in spring and conventional leys in autumn may account for some of the difference observed (Eltun, 1995). A study by Stopes et al. (2002) compared organic farms to conventional farms with similar crop rotations and climate, and found that N leaching was similar or slightly lower from the organic farms when crop yields and inputs were taken into account. Kristensen et al. (1995b) observed no significant differences in the levels of leachable N in soil from conventional and organic farms, but noted that other factors such as crop type; soil type and precipitation made the comparison difficult. A review by Kirchmann and Bergstrom (2001) found that organic farms had lower nitrogen leaching losses, but they also had lower N inputs than conventional farms, so they were not any more efficient in their use of N inputs than conventional farms. The review was unable to compare crop yields and N leaching on organic and conventional farms, which is unfortunate, because crop yields are more relevant to this discussion than N inputs. Webster et al. (2003) showed that under a low input integrated conventional system, N leaching losses were 67 kg N ha^-1 from beans and peas, 46 kg N ha^-1 under winter wheat and barley, and 24 kg N ha^-1 from grass clover leys. Applied N fertiliser rates were 0 kg N ha^-1 for beans and peas, 141 kg N ha^-1 for winter wheat, 137 kg N ha^-1 for winter barley and 68 kg N ha^-1 for grass clover (Webster et al. 2003). This makes the efficiency of fertiliser use in the integrated system for winter wheat (N output/N fertiliser input) 0.49 and for barley 0.62.

Milk yields from grass-clover pasture with no applied N in New Zealand were found to be 83% of those from pastures receiving 400 kg N ha^-1, but the efficiency of N use was much greater when no N fertiliser was applied. At the high rate of N application, only 26% of the fertiliser N and biologically fixed N was recovered in milk and other produce. When the pasture was reliant on biologically fixed N, 52% of the N input was recovered in the farm produce. Applying 400 kg N ha^-1 as fertiliser increased losses of N to the atmosphere and to groundwater by a factor of approximately 3.7-3.9 (Ledgard et al. 1999).

It is difficult to make comparisons between conventional and organic farms because so many conventional farms are not self-contained or self-sufficient units, as organic farms ideally should be, but rather they are often part of a broader national and international food chain. In particular, the separation of arable production and animal production that has resulted from the use of N fertiliser has also affected the efficiency of N use.

Granstedt (1995) examined the N use efficiency of individual conventional and organic farms in Sweden, and the communities that depended on them. Over the whole of Sweden, 86 kg N ha^-1 was lost from farms to denitrification and leaching. In addition, 19 kg N ha^-1 was lost in the wider community as food waste, slaughterhouse waste, domestic waste, and sewage (Granstedt, 1995).

In one area that specialised in intensive animal production, with an average of 0.8 livestock units (LU) per hectare (1 LU is equivalent to a large dairy cow), 101 kg N ha^-1 were imported as fertiliser, 34 kg N ha^-1 were imported as animal feed and 9 kg N ha^-1 were fixed biologically. From this system, 26 kg N ha^-1 were exported as animal products and 30 kg N ha^-1 as crops. Total N losses were 101 kg N ha^-1 from farms and 14 kg N ha^-1 from the wider community. Total N output/N input (N in produce/N inputs) was 0.34 (Granstedt, 1995).

An area with a moderate intensity of animal production (0.5 LU ha^-1) imported 82 kg N ha^-1 as fertiliser, 3 kg N ha^-1 as animal feed and 13 kg N ha^-1 was fixed biologically. From this system, 15 kg N ha^-1 were exported as animal products and 6 kg N ha^-1 as crops. Losses from this system were 97 kg N ha from farms and 96 kg N ha from the community. Total N output/N input was 0.178 (Granstedt, 1995).

An area with a lower intensity of animal production (0.2 LU ha^-1) imported 76 kg N ha^-1 as fertiliser, 6 kg N ha^-1 as animal feed and 5 kg N ha^-1 were fixed biologically. From this system, 6 kg N ha^-1 were exported as animal products and 39 kg N ha^-1 as crops. Losses from this system were 42 kg N ha^-1 from agriculture and 17 kg N ha^-1 from the community. N output/N input was 0.44. In general, N losses were greatest in areas with intensive animal production. (Granstedt, 1995).

In the same study, a mostly arable farm, a dairy farm and a pig production unit were compared to a biodynamic farm at the farm level. The arable system imported 97 kg N ha^-1 as fertiliser, 2 kg N ha^-1 as feed, 7 kg N ha^-1 were fixed biologically, 102 kg N ha^-1 was sold as produce and 19 kg N ha^-1 was lost to leaching and denitrification. The dairy farm imported 91 kg N ha^-1 as fertiliser, 15 kg N ha^-1 as feed, 17 kg N ha^-1 were fixed biologically and 87 kg N ha^-1 was lost to atmosphere and groundwater. The pig unit imported 103 kg N ha^-1 as fertiliser, 135 kg N ha^-1 as feed, 4 kg N ha^-1 were fixed biologically, 93 kg N ha^-1 were sold as animal products and 166 kg N ha^-1 was lost to the atmosphere and groundwater. The biodynamic mixed farm imported no fertiliser, 3 kg N ha^-1 as feed and 50 kg N ha^-1 were fixed biologically, 7 kg N ha^-1 were exported as crops and 14 kg N ha^-1 as animal products. Total losses of N to leaching and denitrification were 45 kg N ha^-1. The biodynamic farm had a yield similar to the Swedish average. Losses of N to leaching were about 25% of the national average, and gaseous losses were 50% of the national average. The biodynamic farm was self-sufficient in nutrients, the other farms were connected in a larger system, and this system had considerable losses of N, because of an imbalance between crop and animal production at both national and local levels. (Granstedt, 1995).

Where cereals are grown for many successive years without a break in the rotation, there is a risk that soil organic matter will be lost (Uhlen, 1991), releasing carbon dioxide into the atmosphere (Pulleman et al. 2000). Soil organic matter supplies essential nutrients and protects them from leaching (Davies, 2000), retains water, raises soil temperature and absorbs cations. It is also needed to maintain soil structure and aeration. Low soil organic matter also makes soils vulnerable to wind erosion. Approximately 2.2 million tonnes of topsoil are lost every year in England and Wales to erosion (Harrod and Fraser, 1999). In the USA, on average 16 tonnes of soil are lost per hectare per year (USDA, 1991). At most 1 tonne of new soil is formed per hectare per year (Pimentel, 1993), which means that losses of N in soil erosion (an estimated 64 kg N ha^-1 year^-1 in the USA) exceed fertiliser inputs (average 48 kg N ha^-1 year^-1, Pimentel, 1996). Forage legumes are particularly good for improving soil structure, because they have a high rate of root turnover, providing substrates for bacteria that produce polysaccharides, important structural components of soil. Decomposing legume roots and mycorrhizal hyphae bind soil aggregates together, increasing their stability (Miller and Jastrow, 1996). A study by Drinkwater et al. (1998), found that legume-based cropping systems retain more soil carbon than those based on nitrate fertiliser. This was attributed to the low C:N ratio of the legume residues, and the greater temporal diversity of the cropping sequences in the organic rotations examined (Drinkwater et al., 1998). There is some disagreement about the effect of mineral N fertiliser on soil organic matter. Thomsen et al. (2003) examined the long term (20-28 years) effects of different N fertiliser applications on spring and autumn sown cereals grown on sandy loam, and concluded that soil organic matter was not increased by high rates of N application. In a 30-year experiment comparing the effects of leys, manure, crop residues and mineral fertilisers, soil carbon content in ley- arable rotations with or without manure remained constant or showed small increases. All cereal rotations with no return of straw showed a 5% decline in soil carbon content (Uhlen, 1991). Kapkiyai et al. (1999) compared different farm management techniques in East African smallholdings. Losses of soil organic matter were greater from N fertiliser-only treatments, than from manure treatments. In contrast, Kidanu et al. (2000) observed that soil organic matter in the Ethiopian highlands was increased by applications of N fertiliser.

Organic farming and other legume-based farming systems are widely believed to pose less of a risk of nitrate leaching than farming systems based on fertiliser. However, the environmental benefits in this regard are less than is commonly imagined. In addition, it should be stressed that the terms "organic" and "conventional" both cover a wide variety of farming systems, so it is difficult to make general statements. The ratio of pasture to arable land, stocking density, and the use of autumn sown crops have considerable impacts on the amount of N leached, and these factors may be more significant than whether legumes, manure or N fertiliser are used. There is some evidence that in the long term, organic farming systems may benefit soil and prevent erosion. Replacing N fertiliser with biologically fixed N could improve the energy efficiency of agriculture, especially in grassland, although organic farming systems have additional energy costs of their own, resulting from mechanical weed control etc., as well as generally lower yields, which can reduce their energy efficiency.

 

4. White clover
1. A description of the plant

White clover (Trifolium repens L.) is a leguminous plant native to the UK. It is a creeping perennial, although its growth and morphology are highly variable:

Seedlings consist of a taproot and stem, from which leaves, petioles, flowers and stolons grow. The stolons branch and form adventitious roots, as well as more leaves, petioles and flowers. After 2 or 3 years of growth, the central taproot senesces, and eventually the stolons become independent plants. Stolons may also become detached from the parent plant by treading or grazing (Caradus, 1990, Frame and Newbould, 1986). Consequently, white clover can spread rapidly, and completely change its pattern of distribution in the space of two years (Thorhallsdottir 1990a,b). White clover produces a variable proportion of "hard" seeds that are slow to germinate, and can remain viable for several years in the soil (Frame and Newbould, 1986). There is some evidence that the degree of hardness in clover seed may be related to the availability of water at the time of seed formation, so it is possible that hard seeds are an adaptation to drought conditions (Thomas, 1987). Clover seeds can pass through the guts of ruminants and subsequently germinate. This may be a significant factor in their dispersal (Young, 2002). White clover morphology varies considerably between varieties. Small leafed varieties, such as Kent wild white, are low growing, have long, branching stolons and are well suited to grazing. Large leafed varieties, such as the Ladino varieties are taller growing, have shorter, less branched stolons and are well suited to cutting for silage (Frame and Newbould, 1986).

2. A history of White clover cultivation

The Romans were aware of the beneficial effects of legumes on soil fertility, and white clover seems to have been held in high regard for many centuries. Homer describes it as "The lotus" and mentions it as fine fodder for horses. In Welsh mythology it is said to have sprung up wherever the Princess Olwen trod, and is celebrated "with praise out of all proportion to its beauty" (Graves, 1948). The deliberate cultivation of white clover probably began in the 13^th century in Moorish Andalucia. Over subsequent centuries it spread first into Lombardy and the Netherlands (both areas under Spanish control), and in 1620, clover seed was exported from the Netherlands to England (for this reason, white clover is sometimes known as Dutch clover). The introduction of white and red clover had wide-ranging consequences: forage and grain production increased dramatically. The potato ceased to be a luxury vegetable and became part of the European staple diet as a result of the improved soil fertility. In Denmark, cattle numbers increased by 1/3 and grain production doubled within 35 years when clover was introduced. The cultivation of clover was a major driving force for the agricultural revolution, and made productive low input agriculture possible (Kjaergaard, 1995).

Today white clover is the most widely grown forage legume in temperate zones, and is especially popular in Europe and in New Zealand. It is also used as a green manure and for undersowing. White clover is an extremely valuable, renewable source of N and animal feed for agriculture. However, during the twentieth century, there has been a tendency for synthetic N fertiliser to replace white clover and other legumes as the source of N for grasslands and agriculture as a whole. The reasons for this have been discussed in section 1.2.7.

 

5. Methods of estimation of Nitrogen fixation
1. Nitrogen benefit, yield and total nitrogen difference methods

The simplest method of estimating nitrogen fixation is to compare yields from legume-based systems receiving no N and legume-free systems receiving varying amounts of fertiliser e.g. organic grass-clover leys and conventional grass-only leys. The fertiliser N equivalent is the amount of N required by the legume free system to match the yield of the legume-based system. One possible source of error in this method is that soil might be enriched under legumes simply because legumes might be less efficient at depleting soil N than non-legumes. This is known as the N sparing effect (Senaratne and Hardarson, 1988). There is also the possibility that N released from the roots of the legume could stimulate the release of additional N from the soil, e.g. by stimulating the mineralisation of soil organic N (Laidlaw et al., 1996).

A more sophisticated version of this approach is to compare the total N content of legume and non-legume biomass:

A non-legume monoculture compared to a mixture of the same non-legume and a legume, grown on the same soil, under identical conditions. Total yield of nitrogen is measured for both plots and species. Assuming that both mixture and monoculture are equally capable of taking up nitrogen from the soil, any additional nitrogen accumulated by the legume mixture will be due to nitrogen fixation (Elgersma et al. 2000).

The legume and non-legume monoculture may not have identical patterns of nitrogen uptake throughout the growing season, and they may not have access to the same sources of soil nitrogen (Bremer et al., 1993). Turkington and Burdon (1983) observed that T. repens and L. perenne (the usual reference crop in ¹⁵N dilution studies of T. repens) have asynchronous growth cycles, and that this explains their ability to cohabit. Grasses such as L. perenne are also known to have a much more extensive and finely branched root system than clover, and longer root hairs. Only 68 % of white clover roots have root hairs, as opposed to 95% in L. perenne. The root hair cylinder (the volume of soil contained within the span of the root hairs around the main root) of L. perenne is far larger than the root cylinder of T. repens. The root hair cylinder of L. perenne has a volume of 411 mm³ per mg of root dry matter. The equivalent figure for T.repens is 68 mm³ mg^-1 (Evans, 1977), and this may allow L. Perenne to exploit soil N more effectively than white clover. Non-nodulating mutant varieties of some legumes exist, and these are the ideal reference plants. Unfortunately non-nodulating white clover varieties are not currently available (Warembourg, 1993). In laboratory studies, legumes can be grown as reference plants in sterile conditions, in the absence of rhizobium. The non-legume monocultures used for estimates of nitrogen transfer should ideally be as close to the mixed plots as possible, because of soil spatial variation (Reichardt et al., 1987).

An even simpler method of estimating N fixation is to assume that nitrogen fixation is proportional to dry matter yield, and simply weigh the legume plants. This method is useful for quick comparisons of large numbers of varieties of a single species of legume, or of rhizobium strains (Hardarson and Danso, 1993). The method assumes that there are no significant variations in pNdfa or N concentration in plant tissues. Correlations have also been found between nodule weight and number, and nitrogen fixation calculated by other methods (Hardarson and Danso, 1993). Carranca et al. (1999) observed significant differences between estimates of N fixation by subterranean clover using this method, and estimates obtained by isotope dilution and natural abundance.

2. Acetylene reduction

Nitrogenase, the enzyme which rhizobium use to convert N₂ into NH₃, also converts acetylene into ethylene (Warembourg, 1993). By incubating legume root nodules or whole root systems in an atmosphere containing acetylene, and measuring the rate at which the acetylene is converted into ethylene, nitrogenase activity can be measured, although the ratio of nitrogenase activity to ethylene production varies between species (Warembourg, 1993). If the total number and mass of root nodules can be measured or estimated, then this method can be used to quantify the total amount of N₂ fixed. Errors may occur because microorganisms other than rhizobium may produce ethylene and the treatment may inhibit the rhizobium activity (Wood, 1995). The technique can only provide short-term measurements of nitrogen fixation. Nitrogen fixation may vary throughout the course of a day. Attempts to extrapolate results obtained this way over the course of a growing season are likely to be inaccurate (Caldwell and Virginia, 1989). The acetylene reduction assay is known to produce significantly different estimates of nitrogen fixation from other methods (Martensson and Ljunggren, 1984).

3. Natural Abundance

Soil N and atmospheric N may naturally contain slightly different concentrations of the isotope ¹⁵N, and this can be utilised to estimate the uptakes of soil and atmospheric N by a legume (a natural abundance technique). The ¹⁵N abundance of available soil N is hard to determine, so usually a comparison with a non-fixing reference plant is made. The ¹⁵N abundance of fixed N is sometimes slightly different from the natural abundance of atmospheric N, because nitrogen fixation preferentially uses certain isotopes. This difference (known as an isotopic fraction effect) can be taken into account when estimates of nitrogen fixation are made (Caldwell and Virginia, 1989). Natural abundance techniques require extremely accurate measurements to be made, because the differences in natural abundance are extremely small, relative to background variation (Handley, 1996). Typically the d ¹⁵N values of clover may vary by less than 1% (e.g. Hanson et al., 2002). Care must be taken to avoid contamination of the samples. Often it is necessary for a mass spectrometer to be used solely for natural abundance, if accurate results are to be obtained (Shearer and Kohl, 1993). Estimates of N-fixation by this method have been known to differ from those obtained by ¹⁵N dilution (Høgh-Jensen and Schjoerring, 1994), although Carranca et al. (1999) observed no significant differences between ¹⁵N dilution and natural abundance methods. This method suffers from the same problems with reference plants as the N difference method.

4. ¹⁵N Dilution

Natural differences between soil and atmospheric nitrogen ¹⁵N abundance values are often small, variable or non existent, so the soil nitrogen pool is often artificially labelled with ¹⁵N. ¹⁵N-labelled fertiliser is added to the soil in which nitrogen fixing and non-fixing reference plants are growing. The uptake of the label is compared in nitrogen fixing and non-fixing species (e.g. Laidlaw et al. 1996). The percentage of N derived from the atmosphere (pNdfa) is calculated from:

[.1]

(Wood, 1996), where %excess¹⁵N_leg is the percentage enrichment of ¹⁵N in the legume as compared to the background (unlabelled) ¹⁵N level in the legume, and %excess¹⁵N_non is the enrichment of the non-legume.

Nitrogen fixation is calculated as:

[.2]

Where %N_leg is the percentage N concentration of legume dry matter and DM is the dry matter yield of the legume. A comparison of label uptake in the non-fixing species, grown alone and in mixture with legumes is used to provide estimates of nitrogen transferred from the legume to the grass (McNeill and Wood, 1990). Total fixation is often assumed to be equal to the amount of atmospheric derived nitrogen in the legume, plus atmospheric nitrogen (N transfer) in the non-legume (Goodman, 1988). Fixed N transferred from the legume to the soil is not measured in this or any other method described here.

This method suffers from much the same problems as the total nitrogen difference and natural abundance methods, with respect to reference plants.

It assumes that both the legume and the non legume must take up soil N in proportion to the amount of soil N available, and that the ratio of fertiliser N:soil mineral N available to both species must be the same. The soil enrichment is likely to decline over time, especially if ¹⁵N is applied as a single application. If enrichment declines with time, then comparisons of legume varieties with different life cycles may not be valid (Danso et al. 1993). Jorgensen et al. (1999) suggested that the method could be improved by immobilising the ¹⁵N using sucrose and straw, as this caused the ¹⁵N enrichment to decline more slowly over the growing season. Errors associated with the reference plant are greatest at low levels of fixation (Danso, et al. 1993). For accurate results from this method, the legume and reference crop should also absorb similar ratios of fertiliser N and soil N during each phase of their life cycle. It is also necessary to assume that the legume and non-legume have identical patterns of nitrogen uptake throughout the growing season, and that both have access to the same sources of soil nitrogen (Bremer et al., 1993). The non-legume monocultures used for estimates of nitrogen transfer should ideally be as close to the mixed plots as possible, because of soil spatial variation (Reichardt et al., 1987).

Usually only herbage above stubble height is sampled (e.g. Høgh-Jensen and Schjoerring, 1994). Sometimes a correction is made for nitrogen present in stolon and root material (Jorgensen and Ledgard, 1997), and this is assumed to be constant, although there seems to be disagreement about this:

Jorgensen and Ledgard (1997) and McNeill and Wood (1990) found that estimates of the proportion of nitrogen derived from the atmosphere (pNdfa) were the same whether roots and stolons were included in the calculations or not. However, estimates of nitrogen transfer are changed if changes in root and stubble N are taken into account (Laidlaw et al., 1996).

There is some phenotypic variation in clover root:shoot dry matter ratios (Burdon, 1980). Fertiliser nitrogen also affects root:shoot ratios (Jorgensen and Ledgard, 1997; Ryle et al. 1981a), as does defoliation (Davidson et al. 1990, Pedersen, 1989); defoliation can also cause an increase in the proportion of soil derived nitrogen in clover roots (Marriott and Haystead, 1992) and a decrease in root and stolon mass (Chapman and Robson, 1992). Legumes are also known to store nitrogen in their root systems over winter (e.g. Li et al., 1996), so root nitrogen may not be constant throughout the growing season. Hay (1985) observed that the percentage of stolon below ground, at the surface and above ground, varied with season: In midsummer, surface and above ground stolon material constituted around 50% of the total stolon weight, but at midwinter, around 90% of the stolon material was below ground. All of these factors suggest that root stolon and stubble sampling could be necessary to obtain accurate measurements of nitrogen fixation.

5. Xylem-solute techniques

These techniques work by measuring the levels of N-containing compounds transported from the roots to the shoots in xylem sap. When this differs between fixing and non-fixing varieties of the same species of legume, it can be used as a measure of nitrogen fixation (Hardarson and Danso, 1993). Sap can be extracted from cut stems using a vacuum pump, or from cut roots, by root pressure. Extracts can also be taken from stems and petioles using hot water (Herridge and Peoples, 1990). This technique is most useful for studying tropical legumes, such as soybean and cowpea, which transport N as ureides, allantoin and allantoic acid. Other legumes transport fixed N as amides, asparagine and glutamine, and show less response to changes in N source in this respect (Hardarson and Danso, 1993). So far this technique has not been used with T. repens.

Measurements of the level of ureide in the sap of soybean plants has been shown to be highly correlated with nitrogen fixation estimates obtained by ¹⁵N labelling, with different strains of soybean and rhizobium (Herridge and Peoples, 1990).

6. 15N₂ Incubation method

If a legume is grown in an atmosphere enriched with ¹⁵N₂, direct measurements of pNdfa can be made. This requires an enclosed, airtight system, which makes this method impractical for use in the field. The volume of the system must be small enough to allow sensitive measurements to be made, and to keep the quantity of ¹⁵N₂ (and therefore the cost) to a minimum (Warembourg, 1993). Problems could arise in a small system if oxygen became depleted, or if CO₂ accumulated at higher levels than would occur in nature. Either of these things could disrupt the normal pattern of plant growth. In long term experiments of this sort, the atmosphere inside the system must be continually circulated and regulated by the addition of oxygen or nitrogen. CO₂ can be removed if necessary by a soda trap. Atmospheric pressure must also be regulated (Warembourg, 1993). Results from this method are similar to those from the ¹⁵N-dilution method and the total N difference method (McNeill et al., 1994).

7. Summary

Advantages and disadvantages of the various methods of measuring N fixation are summarised in Table 1.4.1. Judging by the published evidence, the N difference, ¹⁵N dilution and natural abundance methods are most suitable for measurement of N fixation in the field over long time periods. These three methods can all provide estimates of N transfer from legumes to non-legumes, but the N difference method cannot be used to estimate pNdfa.

 

Method	Time-scale	Sources of error	Can plants be grown in the field?	Ever inconsistent with other methods?	Cost
N difference/ fertiliser equivalent	No time limit	Nitrogen sparing effect 8	yes	Acetylene reduction 6, 15N dilution, Natural abundance 2	Cheap
Acetylene reduction	hours	Ethylene producing bacteria, Inhibition of rhizobium 10	yes	15N dilution and N difference methods 6	Cheap
Natural abundance	No time limit	Background variation, contaminated samples, reference crop 4	yes	15N dilution 5, N difference 2	expensive equipment
15N dilution	No time limit	reference crop, soil variation variable root mass 1, 3, 7,	yes	Acetylene reduction 6, N difference 2	expensive
Ureide assay	short-term	-	yes	-	cheap
15N2 Incubation	<year	atmospheric composition, leaks 9	no	no	expensive

Table .1 Methods of measuring N fixation

References: ¹Bremer et al. (1993) ²Carranca et al. (1999) ³Danso et al. (1993) ⁴Handley (1996) ⁵Høgh-Jensen and Schjoerring (1994) ⁶Martensson and Ljunggren (1984) ⁷Reichardt et al. (1987) ⁸Senaratne and Hardarson (1988) ⁹Warembourg (1993) ¹⁰Wood (1995)

7. Estimates of N fixation by white clover

Estimates of N fixation by white clover have been summarised from available data (Table 1.5.1). Generally, N fixation by white clover is highest when no nitrogen is applied, and declines in proportion to the amount of N fertiliser applied (Figure 1.6.1). Average estimates of N fixation with low N (less than 5 kg N fertiliser ha^-1 yr^-1) are 161.6 kg N ha^-1 in New Zealand, 145.1 kg N ha^-1 yr^-1 in the British Isles, and 130.7 kg N ha^-1 yr^-1 in Denmark. In Switzerland and the Netherlands, the average N fixation values are 215.25 kg N ha^-1 yr^-1 and 359.6 kg N ha^-1 yr^-1 respectively, but these values should be taken with caution as the data for these countries came from a single source each. When only data from ¹⁵N dilution studies is included, average N fixation by white clover is 158.3, 165.7 and 120.6 kg N ha^-1 yr^-1 for New Zealand, Britain and Denmark respectively. There was no clear evidence from this that different methods of estimating N fixation produced different results, but other factors such as N application, and the differences between sites made comparisons different.

 

N fixed (kg ha-1)	Measurement technique	fertilizer(kg ha-1)	Location	Ref.
83.5-171	15N dilution	0	UK	20
82-291	15N dilution	0	NZ	17
82-213	15N dilution	0	NZ	18
40-160	15N dilution	390	NZ	18
27-122	15N dilution	0	UK	7
-89-178	N difference	0	UK	7
158-195	15N dilution	0	Denmark	13
42-200	15N dilution	0	Uruguay	19
0-20	15N dilution	0	U.S.A	10
83-283	15N dilution	0	Switzerland	1
48-173	15N dilution	120	Switzerland	1
165-211	15N dilution	150	Switzerland	1
71.82-109	15N dilution	0	NZ	16
184-232	Acetylene reduction	0	NZ	12
105	15N dilution	0	NZ	26
71-114	15N dilution	3	Denmark	11
54-90	15N dilution	24	Denmark	11
52-85	15N dilution	48	Denmark	11
47-78	15N dilution	72	Denmark	11
150-545	N difference	0	Netherlands	6
114.9-233	15N dilution	1	UK	15
146-167	acetylene reduction	62.5	UK	22
30-50	acetylene reduction	0	UK	2
49	acetylene reduction	0	UK	24
66-81	acetylene reduction	0	Canada	25
83-296	Acetylene reduction	0	Eire	21
76-105	Acetylene reduction	0	NZ	4
211-242	Acetylene reduction	0	NZ	3
45-142	15N dilution	3.6	NZ	5
268	acetylene reduction	0	UK	9
191	N difference	0	Denmark	14
152	N difference	78	Denmark	14
115	N difference	155	Denmark	14
69	N difference	310	Denmark	14
38	N difference	465	Denmark	14
208	N difference	125	Denmark	23
143	N difference	250	Denmark	23
90	N difference	375	Denmark	23
74	N difference	500	Denmark	23

Table .1 N fixation by white clover in different parts of the world and at different levels of N fixation

References: ¹Boller & Nosberger, 1987 ²Bradshaw et al. 1975 ³Clark et al., 1979 ⁴Crush et al. 1983 ⁵Edmeades & Goh, 1978 ⁶Elgersma & Hassink 1997 ⁷Goodman, 1988 9Halliday and Pate, 1976 ¹⁰Heichel & Henjum, 1991 ¹¹Høgh-Jensen & Schjoerring 1997 ¹²Hoglund & Brock, 1978 ¹³Jorgensen et al. 1999 ¹⁴Koefoed & Klausen, 1969 ¹⁵Laidlaw et al., 1996 ¹⁶Ledgard et al. 1987 ¹⁷Ledgard et al., 1990 ¹⁸Ledgard et al., 1996 ¹⁹Mallarino et al. 1990 ²⁰McNeill & Wood, 1990 ²¹Masterson & Murphy, 1976 ²²Palmer & Iverson, 1983 ²³Pedersen & Moller, 1976 ²⁴Skeffington & Bradshaw, 1980 ²⁵Vessey & Patriquin, 1984 ²⁶Wheeler et al. 1997)

 

8. Effect of soil nitrogen on N fixation

As we have seen from section 1.5, N fertilisers significantly reduce N fixation by white clover and other legumes. Figure 1.6.1 gives an indication of the scale of this effect. When N is applied to a legume growing as a monoculture, the legume usually increases its uptake of soil N, and consequently the pNdfa (see section 1.4.4) is reduced. Application of fertiliser N causes a reduction in the number and mass of root nodules (Cowling, 1961). This is probably because uptake of soil N requires less photosynthetic energy than N fixation, and so under conditions of abundant N, the relationship between the rhizobium and its host legume will become increasingly parasitic, rather than mutualistic. The rhizobium is benefiting from its association with the legume, but the legume is not. Similar shifts between mutualism and parasitism have been observed in mycorrhizas (symbiotic plant root fungi, Johnson et al., 1997). Parsons et al. (1993) suggest that the effect of soil nitrogen upon pNdfa, may be caused by a feedback mechanism, in which nodule growth and activity are inhibited by high levels of phloem nitrogen compounds, probably amino acids.

Bergersen et al. (1989) observed that cereal crops grown after ploughing of grassland and before planting of soybeans, reduced available N in the soil. This had the effect of increasing N fixation by the soybeans and also improved yields and protein content of the soybean crop. Davidson and Robson (1986) showed that continuous exposure to low levels of nitrogen reduced pNdfa more than short-term exposure to high concentrations of nitrate. Growing legumes under conditions of high N is not making good use of natural resources. pNdfa will be reduced by high soil N, and there is likely to be a high risk of N leaching especially if the legumes are grown as monocultures (Webster et al. 2003). In organic systems there are also likely to be increased weed problems because of the high availability of N.

When N is applied to mixtures of legumes and non-legumes (such as grass-legume leys), the growth of the non-legume is commonly increased at the expense of the legume (Frame and Newbould, 1986, Danso et al. 1988, Waterer et al., 1994, Stewart and Chestnutt, 1974, Rys and Mytton, 1985, Høgh-Jensen and Schjoerring, 1994 & 1997). This seems to indicate that under conditions of high nitrogen, legumes suffer in competition for light, water, and soil nutrients other than N. Laidlaw and Withers (1998) suggest that in grass-clover swards, applications of N cause grass to shade out clover. Davidson and Robson (1986) observed that under laboratory conditions, high levels of soil N increased grass dry matter and reduced clover dry matter in mixed swards. The N concentration of the clover (N concentration in clover dry matter) was unaffected by N fertiliser, and pNdfa was reduced by N fertiliser inputs indicating that under these conditions, clover was able to compete with grass for soil N. Herrmann et al. (2001) also observed reduced clover yield and pNdfa when N fertiliser was applied at a rate of 80-160 kg N ha^-1. Applications of fertiliser N did not affect the N concentration of the clover herbage (Herrmann et al., 2001). Similar effects have been observed in many other studies e.g. Ledgard et al. (2001), Høgh-Jensen and Schjoerring (1994). In all these examples, N fertiliser caused a reduction in the ratio of clover to grass and a decline in pNdfa. Clover N concentration was in all cases higher than grass N concentration and was unaffected by fertiliser N. This indicates that clover is capable of competing with grass for soil N, although under conditions of high soil N, other competition factors become important.

The effect of N fertiliser on N fixation in ryegrass-white clover swards is shown in Figure 1.6.1. Data was compiled from a number of experiments performed in Europe and New Zealand. There is considerable variation between sites, as would be expected, but overall, N fixation is reduced by approximately 25 kg N ha^-1 for every 100 kg N ha^-1 of fertiliser N applied.

Although N fertiliser generally benefits grass at the expense of clover, it has been suggested that small amounts of nitrogen improve establishment of clover, by delaying nodulation (e.g. Young, 2002). A review of the evidence by Peoples et al. (1995) found that in most cases, even small amounts of N fertiliser suppressed N fixation.

 



Figure .1 Effect of N fertiliser applications on N fixation by white clover-grass leys.

Data from ¹Ledgard et al. 1996, ²Boller and Nosberger, 1987 ³Høgh-Jensen & Schjoerring 1997, ⁴Koefoed & Klausen, 1969, ⁵Pedersen & Moller, 1976

 

9. Effects of grazing and cutting on N fixation in grassland
1. Effects of dung and urine deposition

Grazing removes N from large areas and transfers it to small, localised patches (Schwinning and Parsons, 1996a). Afzal and Adams (1992) found dramatic variations in nitrate and ammonium on cattle grazed pastures, over distances of only 100 mm. This small-scale variation was attributed to individual dung and urine patches. The patchy distribution of nitrogen in grassland caused by grazing could theoretically increase clover growth, by creating low N microsites, in which clover could thrive. Hoglund and Brock (1978) found that grazing depresses nitrogen fixation, probably because of the effects of dung and urine patches (Marriott et al., 1987a) and defoliation of clover (Farnham and George, 1994). Baars and Brands (1996) showed that composted farmyard manure inhibited clover stolon growth less than slurry, but that this effect varied with clover varieties. Jorgensen and Jensen (1996) found no effect of dung upon white clover growth or fixation in the first four weeks after application. This was attributed to the slow release of nitrogen from dung. Vinther (1998) estimated that dung and urine patches reduce N fixation in their immediate vicinity by around 15%. Weeda et al. (1967) observed that clover was more prevalent on cattle dung patches than other species, especially in winter. Clover was quick to colonise dung patches, and remained growing on the patches for approximately 1.5 years after deposition (Weeda et al., 1967). In contrast, Lieth (1960) observed that sites where dung had been deposited, became colonised by species such as Poa trivialis (rough meadow grass), Agrostis alba (wood meadowgrass) and Ranunculus repens (creeping buttercup) at the expense of T. repens, although the study appears to have been much less extensive than that of Weeda et al. (1967). Urine + faeces and urine alone decreased the clover content of swards, but faeces alone might actually benefit the clover (Weeda, 1967). Dung and urine generally contain virtually all of the nutrients necessary for plant growth, in varying quantities. It is probably misleading to think of dung and urine simply as sources of N. The effect of dung and urine on clover growth and N fixation could well depend upon the local soil conditions and the availability of key nutrients.

2. Effects of grassland management on N fixation

Wilde et al. (1983) compared rotational grazing to continuous stocking. After one growing season there was over 50% clover on the rotationally grazed plots, while set stocked plots had 5-7% clover. The study concluded that rotational grazing was better than continuous stocking, because it relieves clover from selective defoliation and allows the clover petioles to elongate, placing leaves in the top of the canopy.

Schils et al. (1999) found that white clover ground cover was greater in a system with cutting but no grazing, than in grazed systems. A rotational system of cutting and grazing reduced clover cover by 12%. The negative effect of grazing was most marked from July and August onwards. Where two silage cuts were taken rather than one, clover cover was on average 8% higher. However, dry matter and N yields of rotationally grazed plots were equal to or higher than those which were cut only, suggesting that return of N in dung and urine compensated for the reduced percentage of clover. Rhodes (1984) states that clover yields from rotationally grazed swards are lower than yields from cut swards. Frame and Newbould (1986) observed that grazing depressed clover content of swards and total production, and that clover performance was greatest in cut swards when the interval between cuttings was increased.

Laidlaw and Stewart (1987) found that maximum clover content of a pasture rotationally grazed by cattle, declined from 55% in the third year, to 24% in the sixth year. Hard grazing by sheep over the next three winters increased the clover content to 34 or 45%, depending upon whether or not nitrogen was added in spring. The corresponding figures for a control plot were 7 and 11%.

Sheep generally preferentially graze clover in grass-clover swards, and this may reduce N fixation. Goats will graze weeds and grass in preference to clover, and can increase the clover content of a sward (Grant et al., 1984).

Trampling and soil compaction caused by grazing animals may also reduce N fixation and clover content of pastures (Curll and Wilkins, 1983).

Acuna and Wilman (1993) noted that clover content in a sward was affected by cutting height: Cutting close to the ground encouraged clover, whereas cutting at 100 mm for several years almost completely eradicated clover. The authors explained the suppression of clover with increased cutting height as a shading effect (clover is a low growing plant compared to many grasses). Another possibility is that close cutting removes more nutrients from the soil than lax cutting, and this could also affect the competition between grass and clover. Experiments comparing regular and infrequent cutting of grass-clover swards, with and without applied N, show that less regular cutting of a grass-clover sward reduces clover yield only when N is applied (Wilman and Fisher, 1996), so it is probably true to conclude that grass often out-competes clover for light, when N is applied, or when the sward is allowed to grow tall. There seems to be a general agreement from all of these studies that cut swards favour clover more than grazed swards, and that clover yields better in rotationally grazed swards than in continuously grazed swards.

3. Effect of companion species

White clover in leys is seldom grown as a monoculture. More usually it is grown with one or more non-leguminous species, or companion species. Ryegrass is generally considered to be a good companion species for white clover, whereas cocksfoot (Dactylis glomerata) is less compatible (Chestnutt and Lowe, 1970). Bent grasses (Agrostis spp.) and Yorkshire fog (Holcus lanatus) are also considered poor companion species for clover (Frame, 1990). In contrast, Edmond (1964) found that the yield of clover grown with Holcus lanatus and Agrostis tenuis was actually slightly higher than yield of clover grown with Lolium perenne. Clover grown with D. glomerata yielded about half that of clover grown with L. perenne (Edmond, 1964)

Williams et al. (2000) compared yields of two medium leaf sized clover varieties (AberDai and AberVantage), grown individually and as mixtures. The clover was grown with four ryegrass varieties. Two of these (Augusta and AberOscar) were tetraploid hybrids between perennial ryegrass and Italian ryegrass (Lolium multiflorium L.) The other two varieties were a diploid Italian ryegrass (AberComo) and a tetraploid perennial ryegrass (Merlinda). The management involved silage cuts and grazing by sheep and cattle each year for four years. Overall there was no benefit from mixing the two clover varieties, but in most years, clover yields were highest with Merlinda (the lowest yielding grass) and lowest with AberComo (the highest yielding grass). This experiment only looked at dry matter yields, and not N yields or N fixation.

Mattner and Parbery (2001) showed that when perennial ryegrass became infected with crown rust, it suppressed the growth of clover more than healthy ryegrass. Soil previously growing rusted ryegrass and leachate from soil growing rusted ryegrass also suppressed clover growth. This was explained as an allelopathic effect.

Competition between clover and grass seems to be generally poorly understood. Competition between clover and other pasture species could depend on light, temperature, water or availability of nutrients, and the ability of the other species in the sward to compete for these resources. A grass variety or species which out-competes clover in the short term is likely to reduce the productivity of the sward in the long term. The species in a sward must effectively utilise soil nutrients to prevent leaching losses. On the other hand, if they compete too aggressively with clover, the productivity of the sward will be reduced in the long term.

4. Temporal variation of N fixation

Hoglund and Brock (1978) noted large differences in nitrogen fixation by white clover with season, and also between years. White clover populations in particular are claimed to "crash" every few years (Fothergill et al., 1996).

Schwinning and Parsons (1996a) suggest that clover and grasses replace each other cyclically at the level of individual patches. Grass has a competitive advantage when soil N is high, and clover is at an advantage when nitrogen levels are low, but clover gradually elevates the level of soil nitrogen locally. They modelled this by representing a pasture as 90,000 interlocking hexagonal cells. These cells could be in one of 4 states: legume dominant, grass dominant (with legume present), pure grass at high soil N and pure grass at low soil N. Cells moved between these states in response to urine deposition, N enrichment by clover, N depletion by grass, invasion by clover and extinction of clover. The output of the model suggested that this would lead to cyclical variation in clover content with a period of 3-4 years (Schwinning and Parsons 1996a). Evidence from field data provided some support for this (Fothergill et al., 1996, Schwinning and Parsons, 1996b). If clover is uniformly distributed, then clover in all parts of the pasture will oscillate in phase. However, clover dies back in winter and must re-invade the areas where it previously occurred, during the rest of the year. This raises the possibility that some patches of clover will disappear altogether, especially when other factors such as grazing, trampling, pests, dung and urine are considered. These processes will increase patchiness. As the pasture becomes more patchy, different regions will no longer be in phase with one another, so oscillations at the field scale will be damped down. Grass yield may be more closely correlated with previous years clover yield than with current clover growth, so that clover cycles and total biomass cycles are not in phase. The amount of dieback is related to climate, and to management (field scale effects) where there is very little dieback, the pasture will take many years to reach equilibrium, a lot of dieback and the pasture reaches equilibrium quickly. (Schwinning and Parsons, 1996a).

Turkington and Harper (1979) suggested the following cyclical succession sequence: T.repens would invade and be joined by L. perenne, the two species would coexist, because of the high nitrogen requirements of L. perenne and the asynchronous growth cycles of the two species. As the soil nitrogen level rose owing to the nitrogen fixation by T.repens, T.repens would go into decline, because of its poor ability to compete for soil nitrogen, and be replaced by Alopecurus pratensis (meadow foxtail) and/or Dactylis glomerata. L. perenne would also decline at this point. The nitrogen level in the soil would then decline and A. pratensis would be replaced by slower growing species with a low nitrogen demand, such as Anthoxanthum odoratum (sweet vernalgrass) and Agrostis capillaris. They also suggest that nitrogen inputs from dung and urine would complicate the picture, by effectively omitting clover from the sequence locally.

Experiments with simulated swards containing a number of grassland species, and observations of old pasture, by Thorhallsdottir (1990a,b) did not support the idea of simple cyclical species replacement, although some patterns of replacement related to T. repens were observed. T.repens was more likely to replace certain species, and be replaced by others, than could be expected from chance. T.repens had a tendency to colonise gaps, and to be replaced by gaps. It also moved rapidly through the pasture, never occupying the same space in successive years more often than would be expected by chance (Thorhallsdottir, 1990a). Cain et al. (1995) observed changes in clover density in a lawn between years, and a general pattern of moving clover patches in a sea of grass. Cyclical replacement of grass species by clover was observed (different grass species were not distinguished). The larger patches of clover persisted between one and three years. Some small patches might have persisted for longer than the 4 years of the study. Lieth (1960) also noted cyclical species replacement in grassland, but failed to describe it in any detail.

Climatic factors such as sunlight, temperature and rainfall also contribute to yearly variation in clover growth (Frame and Newbould, 1984). Hay et al. (1990) note that clover populations consist of a few large individuals, and a large number of small individuals (a result of senescence of stolons). These small individuals could easily die off under harsh environmental conditions, pests, disease etc. Frame and Newbould (1984) also suggest that inappropriate herbicide use might be partly responsible for some clover crashes.

Edmeades and Goh (1978), looked at pastures 2, 6, 15 and >20 years old, and found that nitrogen fixation generally decreased with age of pastures, although this study had no replicates. In contrast, Heichel and Henjum (1991) found that nitrogen fixation in forage legumes increased with the age of the pasture. Some of these differences could have been due to variations from year to year. Høgh-Jensen and Schjoerring (1997) showed that at low seeding densities, clover could take a full growing season to achieve the same yield output as an initially well-seeded sward. Kristensen et al. (1995a) estimated that a pasture containing 30% clover would average 190 kg N ha^-1 yr^-1 in the first two years, and 128 kg N ha^-1 yr^-1 in subsequent years.

5. Effect of temperature on N fixation

At low temperatures, clover is less able to fix nitrogen, and suffers through competition for soil nitrogen with grass (Nesheim and Boller, 1990; Prevost and Bromfield, 1991). Macduff and Dhanoa (1990) also found that temperatures below 13^oC suppress nitrogen fixation. Frame and Newbould (1986) suggest that N fixation by white clover requires a temperature of about 9^o C. At low temperatures, plant growth as a whole is reduced, and so the reduced fixation could simply be a response to low nitrogen demand. Ollerenshaw and Baker (1981) observed that clover roots remained active at temperatures as low as 5^oC. Nitrogen fixation is therefore likely to be affected by seasonal and annual temperature variations, as well as seasonal and annual variations in water availability and soil factors.

 

 

10. Other factors affecting N fixation

Different legume species (Heichel and Henjum, 1991) and cultivars (Ledgard et al., 1990, 1996) fix different amounts of N, and this may be related to their tolerance of soil nitrogen levels (Ledgard et al. 1996).

In the laboratory, water stress reduces nitrogen fixation (Engin and Sprent, 1973), but in the field, drying of the surface soil layers causes nitrogen fixation by T.repens to take place at greater depths, reducing this effect (Hoglund and Brock, 1978).

Topography also seems to be important: sloping sites tend to have lower N fixation, probably because of differences in microclimate and soil fertility (Ledgard et al., 1987), and lowland sites can have four times as much N fixation as upland sites, probably because of the longer growing season (Goodman, 1988). Jacot et al. (2000) observed that at altitudes over 2100m above sea level, T.repens was not present, although Trifolium pratense (red clover) did grow at this altitude and Lotus corniculatus (birdsfoot trefoil) and Trifolium alpinum (alpen klee) could grow at altitudes of 2300m and 2600m respectively. High altitude did not reduce pNdfa values of any of the species in the study, even at the limits of their range, despite low temperatures and acid soils.

A survey of dairy and beef farms by Forbes et al. (1980), showed that clover content of swards was significantly affected by soil drainage, fertiliser N use, whether or not clover was a preferred species, the potential for transpiration from the crop surface and the number of days of drought per year. Soil pH and available soil P and K had no significant effect on clover growth. In contrast, Snaydon (1961) observed that clover distribution in hill pastures was related to levels of Ca and P.

Giller and Cadisch (1995) looked at ways of increasing biological nitrogen inputs to world agriculture. They concluded that in the short term improvements could be made by improving soil conditions, such as acidity, water stress, nutrient deficiencies and high soil nitrogen, using liming, fertiliser, green manure and crop rotation. Simply encouraging the wider use of legumes would also be effective. They commented that: "Immediate dramatic enhancements in input from N₂ fixation are possible simply by implementation of existing technical knowledge". Inoculation of legumes with appropriate rhizobium strains would have immediate benefits in many parts of the world. Breeding of improved legume varieties and rhizobium, and genetic engineering, will only show benefits in the longer term, if at all (Giller and Cadisch, 1995).

Some of these factors have been quantified experimentally, and these are shown in Table 1.8.1. It is clear from this study that much of the research on clover has focussed on the effects of N in fertiliser and in dung and urine, and this seems to cause consistently large reductions in N fixation. Some of the factors, such as clover variety, irrigation and management may not be universally applicable: different clover varieties may perform differently at different sites, and certain management techniques may be more appropriate in some areas than others. Table 1.8.1 only records available data from studies on N fixation. It does not include studies that have recorded factors affecting clover yield only, even though this is likely to have highly significant effects on N fixation (e.g. Acuna and Wilman, 1993). Some of the factors observed to inhibit nitrogen fixation might simply be reducing plant growth as a whole (Hartwig and Nosberger, 1996). Some factors, such as grazing management, are extremely difficult to examine in N fixation studies, because much of the fixed N in clover herbage in a grazed sward is likely to be consumed by the grazing animals.

 

 

 

Factor	Scale of effect (% change)	Reference
Soil nitrogen (16 mg N/plant) establishing sward	94.9%	10
Soil nitrogen (8 mg N/plant) established sward	58.33%	10
Soil nitrogen (465kg ha-1 yr-1)	80.1%	5
Soil nitrogen (400 kg ha-1 yr-1)	57.86%	3
Soil nitrogen (390 kg ha-1 yr-1)	57.6%	6
Soil nitrogen (310 kg ha-1 yr-1)	63.9%	5
Soil nitrogen (155 kg ha-1 yr-1)	39.79%	5
Soil nitrogen (78 kg ha-1 yr-1)	20.42%	5
Soil nitrogen (72 kg ha-1 yr-1)	27.7%	4
Soil nitrogen (48 kg ha-1 yr-1)	18.1%	4
Soil nitrogen (24 kg ha-1 yr-1)	14.5%	4
4 cuts (compared to 3)	-41%	8
5 cuts (compared to 3)	-46%	8
Sandy soil (in comparison with clay soil)	7.7%	2
3rd/4th/5th years pasture (compared to 1st & 2nd years)	34.6%	2
Irrigation	-15.4%	2
Clover variety (Kopu vs Sabeda)	60.8%	6
Pattern of nitrogen supply	18.79%	1
Waterlogging	97%	9
Temperature (5-15oC)	98.35%	7
Dung and urine patches	10-15%	11

Table .1 Factors reducing nitrogen fixation in white clover leys

References: ¹Davidson and Robson (1986) ²Gregersen (1980) ³Høgh-Jensen and Schjoerring (1994) ⁴Høgh-Jensen & Schjoerring (1997) ⁵Koefoed & Klausen (1969) ⁶Ledgard et al. (1996) ⁷Nesheim and Boller (1990) ⁸Pedersen and Moller (1976) ⁹Pugh et al. (1995) ¹⁰Rys & Mytton (1985) ¹¹Vinther (1998)

 

 

11. Summary

Like any plant species, white clover has a niche, a set of environmental conditions, in which it can grow and compete with other plants for soil nutrients, water and sunlight. The fact that white clover is found growing in so many different parts of the world, and habitats, suggests that it has a fairly broad niche, and some varieties may be adapted to local conditions. Although white clover has been cultivated and bred, the cultivated forms still resemble those that are found in the wild, and the cultivated varieties can themselves grow in the wild. This is probably because many of the grasslands, in which clover is grown could be described as semi-natural habitats. Clover has the ability to fix nitrogen and spreads rapidly, and this suggests that it may sometimes behave in the wild as a pioneer species, rapidly colonising disturbed and bare ground. Pioneer species are generally transient, and replaced by other species in a successional cycle. This may explain why clover yields can fluctuate from year to year. The fact that clover elevates soil N while at the same time being vulnerable to high soil N, may also affect its growth and N fixation over time.

In order for clover to yield well, the conditions on the farm must resemble to some degree the conditions for which clover has evolved. In addition, for clover to fix nitrogen effectively, it requires a good supply of all nutrients besides nitrogen. A surplus of soil nitrogen reduces the ability of clover to compete with grass. Competition between clover and grass under these circumstances must depend on factors other than N. In Britain, there is some evidence that one of the most important factors affecting the balance of competition between clover and grass is light. Grass species such as

L. perenne grow taller than clover, and under high N conditions are able to grow rapidly and shade out clover. Close grazing or cutting of grass clover swards could reduce the shading of clover by grass, and thus mitigate the effects of high soil N. Temperature and management may also interact with soil N. Clover grows less well than grass at low temperatures and so a high level of soil N in spring and autumn could severely suppress the growth of clover at a time when it is vulnerable to competition from grass.

3. Aims of study

The overall aims of this study were to quantify the extent and scale of variation of nitrogen fixation in organic white clover/ryegrass leys, and to identify the causes of this variation. The study aims to measure and compare the effects of season, crop rotation, soil N and competition on N fixation and growth of white clover. In particular, the study aims to test a number of hypotheses about the relationship between availability of soil nitrogen and grass- clover competition, which will be described in section 2.1.

1. Specific objectives
1. Cyclical Replacement

2. N cycle in grass/white clover leys
1. Experimental sites

This part of the study consisted of two main field experiments: a study of N fixation and N transfer by white clover and N uptake by grass (Section 4 N fixation and grass-clover dynamics) and a study of soil fertility (Section 5 Soil chemistry of grass-clover swards). The field experiments were carried out on two experimental organic farms managed by the Scottish Agricultural College. The two farms contained experimental ley-arable rotations, which had been established several years previously in order to compare the effects of rotations with different ratios of grass-clover ley to arable crops. Although the two farms in this study are in the same broad region (North-Eastern Scotland), they have distinctly different climate and soil fertility.

1. The Tulloch Organic Unit

The Tulloch organic farm is located at the SAC’s Craibstone Estate, Aberdeen (Latitude N 57^o 10’ Longitude W2 ^o14’, National Grid reference NJ843094). The farm covers 65.8 ha of exposed, marginal land, 160 m above sea level, of which 21.9 ha are arable land, 37.7 ha are permanent grazing and 6.2 ha are trees and buildings. The soil type is a sandy loam of the Countesswells series (leptic podzol in FAO classification). In 1997 at the start of sampling, soil pH averaged 5.8 and the soil contained 9.5% organic matter. Soil nutrient levels were moderate to high. P, K and Mg levels in extracts were 15, 97 and 88 mg l^-1, respectively (M.Coutts pers. comm.). In 1992, the Soil Association certified the farm organic.

A rotational trial was established on the farm in 1991. The trial comprises two rotations each replicated twice. Plots are 26 ´ 30m (0.078 ha and the layout is shown in Figure 3.1.1.



Figure .1 Layout of the trial rotation plots at Tulloch in 2000.

The rotations are referred to as the 66% ley rotation and the 50% ley rotation. The cropping sequences are shown in Table 3.1.1

	Cropping sequence
66% ley	Ley (grazed)	Ley (cut)	Ley (grazed)	Ley (cut)	Oats	Oats (u/s)
50% ley	Ley (grazed)	Ley (cut)	Ley (cut)	Oats	Swedes	Oats (u/s)

Table .1 Cropping sequences for the 66% ley rotation and the 50% ley rotation at Tulloch. u/s = undersown with grass and clover

The final year of oats in each rotation is undersown with perennial ryegrass (var. Condessa), Timothy (var. Scots) and White clover (var. Avoca). Ley plots are grazed rotationally by sheep, stocked at a rate of 1.7 livestock units per forage hectare. Farmyard manure (FYM) from organic livestock on the Tulloch unit is applied to the trial. Manure applications are based on the area of forage in the rotation, an assumed 7.2 tonnes manure available per livestock unit. FYM is only applied to cut leys. FYM is applied in early spring and immediately after the first silage cut when a second silage cut is to be taken. Dates of silage cuts are shown in Table 3.1.2. Application rates for FYM are shown in Table 3.1.3.

Age of ley	1997	1998	1999	2000
2 (1st cut)	23.6.97	30.6.98	30.6.99	19.6.00
2 (2nd cut)	9.9.97	21.9.98	5.11.99	29.8.00
3	23.6.97	30.6.98	30.6.99	19.6.00
4	23.6.97	30.6.98	30.6.99	19.6.00

Table .2 Dates of silage cuts on leys at Tulloch, 1997-2000

Age of ley	rotation	1997	1998	1999	2000
2 (1st cut)	66%	20	20	15	15
	50%	15	15	10	15
2 (2nd cut)	66%	10	10	8	8
	50%	8	8	6	8
3	50%	15	15	10	10
4	66%	20	20	15	15

Table .3 Manure applications to cut leys at Tulloch 1997-2000

Management is in accordance with organic standards (Soil Association, 2000a). The trial rotations are in effect a "farm within a farm", and provide a good opportunity to measure N flows under controlled organic conditions. At the time of the start of the study, the trial rotations had undergone one complete cycle, which means that all of the plots in each rotation had had the same number of seasons of ley and arable treatments.

Weather data for Craibstone is shown in Figure 3.1.2, Figure 3.1.3, Figure 3.1.4 and Figure 3.1.5. The growing season in 2000 was relatively cool compared to other years (Figure 3.1.5). 1999 was dry compared to other years (Figure 3.1.4). Generally the peak of temperature was in August, although in 1999 the warmest month was July, and in 1998 it was September. Rainfall showed no clear seasonal pattern.



Figure .2 Average monthly soil temperature and total monthly rainfall in 1997.



Figure .3 Average monthly soil temperature and total monthly rainfall in 1998



Figure .4 Average monthly soil temperature and total monthly rainfall in 1999



Figure .5 Average monthly soil temperature and total monthly rainfall in 2000

2. The Aldroughty Organic Unit

The Aldroughty organic farm is located near Elgin (Latitude N57:38 Longitude W3:23, map reference NJ167625) and covers 56.7 ha of which 43.1 ha are arable land and 13.6 ha are permanent grazing. The site is on sheltered land 25 m above sea level. The soil type is loamy sand/sandy loam. The farm was certified organic by the soil association in 1992.

The farm includes two organic trial rotations, one of which is a replicate of the 50% ley rotation at Tulloch. The plots are 26 ´ 30m (0.078 ha) as in the Tulloch trials, and the leys are managed in a similar way to those at Tulloch. The rotations will be referred to as the 38% ley rotation and the 50% ley rotation. The cropping sequences are shown in Table 3.1.4. Precise weather data for Aldroughty was not available for the period of the experiment.

 

	Cropping sequence
38%	Ley	Ley	Oats	Roots	Oats u/s	Red clover	Roots	Oats u/s
50%	ley	ley	ley	oats	Roots	Oats u/s

Table .4 Cropping sequences for the 66% ley rotation and the 50% ley rotation at Aldroughty. u/s = undersown with grass and clover

3. N fixation and grass and clover dynamics

This part of the study is concerned with the growth, chemical composition, morphology and yield of clover and grass in different years, ages of ley and rotation. Observations of the relative yields of clover and grass (and their changes over time) give a general indication of the balance of competition between clover and grass. Low total yields of all species, may indicate that some external factor e.g. the weather, is affecting the growth of both species. The isotope dilution experiment allows the quantification of the amount of fixed N present in clover herbage (pNdfa), and in the process provides figures for N concentration of clover herbage and grass herbage. This can give an indication of the importance of N in the grass- clover dynamics. For example, if pNdfa was low, clover N yield was high relative to grass N yield, and grass N concentration was high, this would indicate that clover was successfully competing for soil N. A high pNdfa, high clover N yield, and low grass N concentration would indicate that soil N is limiting grass growth. A low pNdfa, low clover N yield and high grass N concentration would indicate that N is abundant and not limiting the growth of grass, with the result that grass is suppressing the growth of clover due to competition for some other factor. Isotope dilution can also indicate how much N is being transferred from clover to grass over the course of the experiment.

From existing work at Tulloch it was known that clover content of the sward in autumn generally varied between approximately 15 and 45% ground cover (Figure 4.5.22). It is not known to what extent this reflects variation in N fixation, as measurements of pNdfa and N concentration have not been made. There have been few studies looking at N fixation over successive years of a ley arable rotation. Data on the flowering rates, and tap-root disappearance in white clover are lacking, but there is some evidence that flowering is influenced by the availability of light and nutrients (Zaleski, 1970). Flowering rate may be inversely related to persistence in white clover varieties (Williams, 1987). Early disappearance of the tap-root can be prevented by lax grazing and applications of P and K (Westbrook and Tesar, 1955).

1. Aims

The overall aim of this part of the study was to obtain information about N fixation by white clover in leys, and its relationship with age of ley, competition with grass, crop rotation, soil factors (especially N) and climate.

It was expected that the past history of a plot would influence the level of N fixation. Older leys were expected to have lower N fixation levels than younger leys, because of the accumulation of fixed soil N in the soil over the previous years of ley. Likewise, it was expected that the 66% ley rotation would have lower N fixation than the 50% ley rotation, owing to the greater amount of accumulated soil N resulting from the extra year of ley, and the shorter N depleting arable phase of the 66% ley rotation.

In order to do this, the following factors were examined: clover and grass yield, N concentration in herbage of clover and grass, pNdfa, N fixation and clover flowering rate.

 

2. Hypotheses

1. Methods
1. N fixation using ¹⁵N dilution technique

1997

In 1997, only the 1-year-old leys at Tulloch were studied (plots 3,10,13 and 19). In each plot, three subplots were chosen, by randomly placing quadrats on the ground. Subplots were covered with exclusion cages (converted lobster pots), to prevent grazing. Three more subplots were chosen in clover-free areas as controls. Subplots were 0.5 m x 0.5 m (0.25 m²). The subplots were cut down to ground level using shears, at the start of the experiment (initial sample) and the herbage was separated into grass and clover. Herbage was separated into grass + weeds and clover, and any soil removed. In a few instances, sheep managed to damage the cages, and partially graze the subplots. When this happened, it was noted and the dry matter yields of these subplots on these dates were not included in the final analysis. The fresh samples were placed in uniform paper bags and weighed, zeroing the balance with an empty paper bag. Samples were then dried overnight at 80 ^oC, along with an empty paper bag, before weighing again, this time zeroing the scales with the dried paper bag. The dried samples were then ball milled and analysed for their N and ¹⁵N concentration on a mass spectrometer (Europa Scientific Tracermass stable isotope analyser).

The standard used to calibrate the mass spectrometer was 2.3584 g of (NH₄)₂(SO₄) (21.21%N, atom% 0.36600), dissolved in 25 ml of deionised water, equivalent to 4.7168 m g in 5m l. Immediately after removal of the herbage, ¹⁵N labelled fertiliser was applied to each subplot. The fertiliser was prepared as follows:

For each of the 15 subplots, 589.29 mg of (¹⁵NH₄)₂(SO₄) were carefully weighed and mixed with 2 l deionised water in a plastic screw top bottle, and another plastic bottle was prepared with 2 l of deionised water. The (¹⁵NH₄)₂(SO₄) had an isotopic enrichment of 10%. In the field, each subplot was watered with 2 l of the ¹⁵N mixture from a watering can and subsequently 2 l of deionised water from a separate and clearly marked watering can. This gave an application rate of 5 kg N ha^-1 and 0.5 kg ¹⁵N

ha^-1. A low application rate of highly labelled ¹⁵N was used because the sward would not normally receive any artificial N fertiliser. The cutting and labelling procedure was repeated every 28 days, until the end of the growing season. Owing to delays in the delivery of the (¹⁵NH₄)₂(SO₄), the initial sample was not taken until 3.7.1997, and only four subsequent cuts were taken. The final labelling was made after the herbage was cut on 23.9.1997, and the final herbage sample was cut on 21.10.1997. Sampling dates for Tulloch are shown in Table 4.3.1.

1998

In 1998, 1 -year-old leys (plots 2, 9, 18 and 24) and 2-year-old leys (3, 10, 13 and 19) were studied (Figure 3.1.1). Three new subplots within each plot were chosen at random. The initial sample of the 1-year-old leys was taken on 29.05.1998, and the initial sample of the 2-year-old leys was taken on 26.05.1997. ¹⁵N label was applied to each subplot and herbage was separated into grass and clover, dried and analysed as described previously. Samples were taken every 28 days from these dates, until 28.10.1997, giving a total of 5 sampling dates for each plot. The final ¹⁵N applications were performed on 29.09.1997 and 22.9.1997 for the 1-year-old and 2-year-old leys, respectively (Table 4.3.1). Three new control plots for the 1-year-old leys and three new control plots for the 2-year-old leys were chosen in locations that were naturally free of clover. The 2-year-old leys were being cut twice for silage during the course of the experiment. When this happened, the cages were removed from the 2-year-old leys, and replaced with marker canes. During silage cutting, the subplots were covered with plastic sheeting, firmly staked down at the corners. This ensured that the grass and clover in the subplots was not damaged, and also prevented grass clippings from elsewhere in the ley, or manure (spread immediately after silage cutting) from contaminating the subplots.

1999

In 1998, 1,2 and 3-year-old leys were studied as well as 1-year-old leys at Aldroughty. As before, three new subplots were selected within each plot. Three control subplots were chosen for the 1-year-old leys, 2-year-old leys, and 3-year-old leys at Tulloch, and the 1-year-old leys at Aldroughty, making 12 clover-free control subplots in all. The controls were placed in nearby grass-clover leys of the appropriate age. Unlike previous years, the controls were not placed in clover-free zones. Instead, the clover present in the subplots at the start of the experiment was removed by hand. Labelling and analysis followed the same procedure as the previous year, with five applications of label and six sampling dates. Sampling dates are shown in Table 4.3.1 and Table 4.3.2. During the silage cutting, subplots in cut leys were covered with plastic as in the previous year.

2000

In 2000, 1-, 2-, 3- and 4-year-old leys were sampled at Tulloch and 1-year-old leys at Aldroughty. At Tulloch, three new control subplots were chosen within each of plots 16, 18, and 13 for 1, 3 and 4-year old leys, respectively and close to plot 17 for 2-year-old leys. The clover in these plots was removed by hand and lawn edging was used to prevent clover from re-invading. The control plots for Aldroughty were created in a nearby ley. Labelling, sampling and analysis followed the same procedures as the previous two years. As before, subplots in cut leys were covered with plastic during silage cutting. Sampling dates are shown in Table 4.3.1 and Table 4.3.2.

Age of ley	Sample	Sampling dates
		1997	1998	1999	2000
1	Initial	7th July	9th June	31st May	22nd May
	1	-	7th July	28th June	19th June
	2	29th July	4th Aug	26th July	17th July
	3	26th Aug	1st Sept	23rd Aug	14th Aug
	4	23rd Sept	29th Sept	20th Sept	11th Sept
	5	21st Oct	27th Oct	18th Oct	10th Oct
2	Initial		2nd June	3rd June	25th May
	1		30th June	1st July	22nd June
	2		28th July	29th July	20th July
	3		25th Aug	26th Aug	17th Aug
	4		22nd Sept	23rd Sept	14th Sept
	5		20th Oct	21st Oct	12th Oct
3	Initial			7th June	29th May
	1			5th July	26th June
	2			2nd Aug	24th July
	3			30th Aug	21st Aug
	4			28th Sept	18th Sept
	5			25th Oct	16th Oct
4	Initial				1st June
	1				29th June
	2				27th July
	3				24th Aug
	4				21st Sept
	5				19th Oct

Table .1 Sampling dates for ¹⁵N dilution study of leys of different ages at Tulloch 1997-2000

Age of ley	sample	Sampling dates
		1999	2000
1	initial	June 11th	May 29th
	1	July 6th	June 27th
	2	Aug 5th	July 24th
	3	Sept 2nd	Aug 21st
	4	Sept 30th	Sept 15th
	5	Oct 28th	Oct 12th

Table .2 Sampling dates for ¹⁵N dilution study of leys of different ages at Tulloch 1997-2000

2. Flowering rate

In 2001, numbers of white clover flower-heads were counted in the herbage samples that had been collected for the ¹⁵N dilution study. The reason for this was that unexplained variations in clover N concentration had been observed in previous years (See section 4.5.3), and it was hypothesised that these may have been due to differences in the proportions of clover dry matter allocated to leaves, petioles, stolons and flowers. Flower buds were counted if they had emerged from the stipule, as were ripe seed heads. Flower stalks were not counted if they had lost the seed head, and detached seed heads were also not counted. Flowerheads were counted to try to explain variations in the N concentration of clover herbage that had been observed previously, and also to test hypotheses about changes in morphology of clover plants in response to soil N.

Flowering rate of white clover was calculated by the formula:

Flowering rate of clover = number of flowerheads per sample/dry matter of clover in sample. [.1]

Flowering rate was analysed by ANOVA on GENSTAT version 6. Model terms were: age of ley (1-, 2-, 3- or 4 years old), date (calculated as number of days after May 1st) and rotation (50% ley or 66% ley).

3. Tap-roots

In August 2001, soil samples collected from clover microsites for chemical analysis (see section on soil chemistry for sampling methods), were examined for clover tap-roots. The number of tap-roots in the bulk sample of soil from clover microsites was recorded for each plot. This gave a figure for the number of tap-roots in 8 soil cores from clover microsites.

2. Calculations
1. pNdfa

Proportion of N derived from the atmosphere, including transfer (pNdfaIT) was calculated from:

[.1]

where %XS¹⁵NC is the ¹⁵N enrichment of the clover sample, calculated as the difference between the percentage ¹⁵N content of the sample and the percentage ¹⁵N content of the initial clover sample (background level) from that subplot. Likewise, %XS¹⁵NG_mo is the ¹⁵N enrichment of the grass monoculture, calculated as the difference between the percentage ¹⁵N content of the grass monoculture sample and the percentage ¹⁵N content of the initial sample from the grass monoculture, averaged for all the monoculture plots of a particular age.

Proportion of N derived from the atmosphere, excluding transfer (pNdfaET), was calculated from:

[.2]

Where %XS¹⁵NG_mx is the ¹⁵N enrichment of the grass in the grass-clover mixture, calculated as the difference between the ¹⁵N content of the grass in the mixed sample and the ¹⁵N content of the grass in the initial sample from that subplot.

2. N fixation

N fixation including transfer (NfixIT) was calculated from:

[.3]

Where DMC is the dry matter yield of clover (kg ha^-1) and %NC is the percentage N (by weight) in clover dry matter.

N fixation excluding transfer (NfixET) was calculated from:

[.4]

To account for fixed nitrogen present in roots and stolons, Jorgensen and Ledgard, (1997) suggest multiplying NfixET by 1.7. Data for NfixET has been presented without this multiplication factor, except where stated.

N fixation was also calculated by the N difference method:

[.5]

Where %NG_mx and %NG_mo are the %N in the grass mixture and the grass monoculture, respectively and DMG_mx and DMG_mo are the dry matter yields of the grass mixture and the grass monoculture, respectively.

3. N transfer

N transfered from clover to grass was calculated by two isotope methods:

a) [.6]

and

b)[.7]

(Farnham and George, 1994)

This method takes account of the possibility that N transfer from clover to grass might increase uptake of soil N by grass, and increase the yield and nitrogen concentration of the grass.

N transfer was also calculated by a comparison of nitrogen yields in grass mixtures and monocultures (referred to as the N difference method):

[.8]

The calculation treats non-fixed N in clover as if it were part of the grass fraction (Ledgard, 1991).

4. Soil derived N

Soil derived N in herbage/kg ha^-1 = (N yield of grass/kg ha^-1 + N yield of clover/kg ha^-1 – NfixET/kg ha^-1) [.9] (modified from Danso et al. 1988).

Soil derived N in herbage (SDN) can be used as an estimate of plant available soil N. This assumes that clover, grass and weeds have the same capacity to utilise soil N. A similar approach was used by Doyle et al. (1986) to estimate available soil N. SDN has two components: Grass N yield and soil derived N in clover. The relative proportions of these two fractions give an indication of the ability of clover to compete with grass for soil N.

Data were analysed by Residual maximum likelihood (REML) and ANOVA on Genstat version 6. Model terms were: age of ley (1-, 2-, 3- or 4 years old), sampling period (1, 2, 3, 4 or 5), year (1997, 1998, 1999 or 2000), and rotation (50% ley or 66% ley). For some of the analysis it was necessary to combine age of ley and year into a single variable (1997(1), 1998(1), 1998(2), 1999(1), 1999(2), 1999(3), 2000(1), 2000(2), 2000(3)), because of the unbalanced nature of the data. 4-year-old leys were examined separately in an analysis of the 66% ley rotation in 2000, because the 50% ley rotation did not have a fourth year of ley. Response variates were: Proportion of N derived from the atmosphere, excluding N transfer (pNdfaET), N fixation excluding N transfer (NfixET), %N concentration in clover dry matter, %N concentration in grass dry matter, dry matter of clover, dry matter of grass and soil derived (non-fixed) N in grass and clover. Because of non-normal residuals in the REML analysis, and a few values slightly higher than 1 (probably due to error in the measurements), pNdfaET was transformed by the following formula:

[.10]

N fixation values also had non-normal residuals and were transformed by the formula:

[.11]

and clover dry matter was also transformed to:

[.12]

Dry matter of grass and soil derived N were both transformed by logging. %NC and %NG were not transformed.

It was not possible in this experiment to examine the effects of cutting or grazing on N fixation, because the grass and clover in the subplots was all being cut at monthly intervals, regardless of the management of the surrounding leys. However, grazing generally began earlier in the spring than the ¹⁵N dilution experiments, so early spring grazing may have affected the subplots in grazed plots. The level of statistical significance in the REML analysis was set at p=0.05.

3. Results
1. Nitrogen fixation

Tulloch

Nitrogen fixation was largely a function of clover dry matter, although this relationship varied slightly between years and ages of ley, owing to variations in the pNdfa and N concentration of the clover (sections 4.5.2 and 4.5.3). The lowest N fixation per unit dry matter was observed in 1999 (Figure 4.5.1, Figure 4.5.2, Figure 4.5.3 and Figure 4.5.4).



Figure .1 Nitrogen fixation plotted against clover dry matter yield in 1-year-old leys at Tulloch in 1997. Each data point represents data from one subplot.



Figure .2 Nitrogen fixation plotted against clover dry matter yield in 1- and 2-year-old leys at Tulloch in 1998. Each data point represents data from one subplot.

 



Figure .3 Nitrogen fixation plotted against clover dry matter yield in 1-, 2- and 3-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 1999. Each data point represents data from one subplot.

 



Figure .4 Nitrogen fixation plotted against clover dry matter yield in 1-, 2-, 3- and 4-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 2000. Each data point represents data from one subplot.

Data from all years and ages of ley was included in a single REML analysis. N fixation varied significantly between different years and/or ages of ley (REML p<0.001). N fixation followed a general seasonal trend, peaking roughly in July and August, and declining in September and October (REML p<0.001), and this pattern varied between different years and/or ages of ley (REML p<0.001). Overall, there was no effect of rotation on N fixation, but in some years and/or ages of ley the two rotations had significantly different N fixation values (REML p<0.001). In order to understand this variation better, the analysis was repeated on each year and age of ley individually.

In 1997 NfixET was highest on the first sampling date (7th-29th July) and subsequently declined (Figure 4.5.5, REML p<0.001). In 1997 1-year-old leys, NfixET was similar in both rotations, and there were no differences between the seasonal patterns of N fixation in the two rotations.

Total N fixation was 41.84 kg N ha^-1. Taking into account fixed N present in roots, stolons and stubble, the total N fixed between July and October can be estimated at 71.55 kg N ha^-1.

N fixation was not measured in June of 1997, but it is possible to estimate the value for June based on data for other years and ages of ley. There is also likely to have been some N fixation in May, although this was not measured directly in any year, because the ¹⁵N label was not applied until the end of June.

In other years and ages of ley the N fixation in June was 22–23% of the total N fixation for the sampling period. Assuming a similar seasonal pattern in 1997, the N fixation in June 1997 can be estimated at 16.1 kg N ha^-1.

The N yield of clover herbage harvested from ungrazed leys prior to labelling (the initial sample) in 1998 2-year-old leys was 8.76 kg ha^-1 (27% of the total for the growing season). In 1999 it was 26.12 kg ha^-1 for 2-year-old leys (33% of the total) and 22.79 kg ha^-1 for 3-year-old silage plots (23% of the total). In 2000 it was 30.19 kg ha^-1 for 2-year-old leys (39% of the total), 23.78 kg ha^-1 for 3-year-old plots (29% of the total) and 21.49 kg ha^-1 for 4-year-old leys (58% of the total).

Assuming that pNdfa of clover was similar in May and June, and that root:shoot ratios were also similar in both months, the total N fixation figures should be multiplied by a factor of c.1.3 to provide a more realistic estimate of annual N fixation, including N fixed outside the sampling period. This means that, in 1997, N fixation in 1-year-old leys may have been as high as 114 kg N ha^-1. N transfer also needs to be taken into account (see section 4.5.11).

 

 

 



Figure .5 Nitrogen fixation excluding N transfer at Tulloch in 1-year-old leys in 1997. Values represent the accumulated N fixed during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1998 NfixET showed a clear seasonal pattern, declining towards the end of the growing season (REML p<0.001). The seasonal pattern varied with age of ley, showing a clear peak in yield between late July and early September in the 2-year-old leys but not in the 1-year-old leys (REML p<0.001, Figure 4.5.6). N fixation in 1998 was significantly lower than in other years. Total NfixET above ground during the period of sampling was 26.2 and 30.31 kg N ha^-1 for 1- and 2-year-old leys, respectively. When N fixation below ground and early in the growing season is taken into account, the estimates became 57.9 and 66.96 kg N ha^-1 yr^-1 for 1- and 2-year-old leys, respectively. N fixation was not significantly different in the two ages of ley. The two rotations also had similar N fixation, and there were no indirect effects of the rotations on N fixation.



Figure .6 Nitrogen fixation excluding N transfer at Tulloch in 1998, 1- and 2-year-old leys. Values represent the accumulated N fixed during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1999, N fixation peaked between July and September (REML p<0.001), but the seasonal pattern was different in different ages of ley, with 3-year-old leys peaking in July and 1-year-old leys fixing most N in late August/early September (REML p<0.001, Figure 4.5.7). Overall, N fixation was highest in 2-year-old leys (REML p<0.001). N fixation over the sampling period was 59.18, 75.64 and 47.63 kg N ha^-1 in 1-, 2- and 3-year-old leys, respectively. Estimated total annual N fixation, above and below ground, was 131.56, 168.15 and 105.88 kg N ha^-1 yr^-1 for 1-, 2- and 3-year-old leys, respectively. N fixation in 1999 was not affected significantly by the type of rotation.



Figure .7 Nitrogen fixation excluding N transfer in 1999, 1-, 2- and 3-year-old leys (Tulloch) and 1-year-old leys (Aldroughty). Values represent the accumulated N fixed during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 2000 N fixation once again showed a strong seasonal trend, declining sharply in autumn (REML p<0.001, Figure 4.5.8). When 1–3-year-old leys from both rotations were considered, the rotation had no significant effect; however, 1 and 3-year-old leys in the 66% ley rotation had significantly higher N fixation than 3-year-old leys in the 50% ley rotation (REML p<0.001, Figure 4.5.9). N fixation was similar in 1,2 and 3-year-old leys (73.7, 70.4, 69.3 kg N ha^-1 respectively) but significantly lower in 4-year-old leys (33.5 kg N ha^-1) REML p<0.001, Figure 4.5.8). Allowing for N fixed below stubble height and early in the growing season, these figures become 163.8, 156.5, 154.1 and 74.5 kg N ha^-1 for 1-2-3- and 4-year-old leys, respectively

When 1–4-year-old leys in the 66% ley rotation only were compared, 3 year-old leys in the 66% ley rotation were seen to have a significantly different seasonal pattern from 1,2 and 4-year-old leys in the 66% ley rotation. N fixation in 3-year-old leys in the 66% ley rotation was highest in the first sample (late May–June), whereas the 1,2 and 4-year-old leys in this rotation showed a peak of N fixation in the 2nd-4th samples (late June-Mid September, REML p<0.001, Figure 4.5.9).



Figure .8 Nitrogen fixation excluding N transfer in 2000, 1-, 2-, 3- and 4-year-old leys (Tulloch) and 1-year-old leys (Aldroughty). Values represent the accumulated N fixed during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

 

In order to examine differences in N fixation between years, 1-year-old leys in 1997, 1998, 1999 and 2000 were compared. N fixation varied significantly with year, lowest in 1998 and highest in 2000 (REML p<0.001). In all years, N fixation declined in autumn (REML p<0.001, Figure 4.5.5-Figure 4.5.8) in 1-year-old leys, but the seasonal pattern varied significantly from year to year (REML p<0.001). In 1999, N fixation was relatively low early in the growing season, but in 2000, N fixation was highest at the start of the season in 1-year-old leys. There were no effects of rotation on 1-year-old leys overall, but there was a significant year ´ month ´ rotation interaction (REML p<0.005) because N fixation was higher in the 66% ley rotation in early and mid-growing season in 1998, 1999 and 2000.

The same procedure was repeated for 2-year-old leys in 1998, 1999 and 2000; and 3-year-old leys in 1999 and 2000. In 2-year-old leys, there were significant differences between years (REML p<0.001) probably due to the low N fixation in 1998, and also significant seasonal variation similar to that seen in the 1-year-old leys (REML p<0.001). The seasonal patterns also varied between years (REML p<0.001), in a similar manner to the 1-year-old leys. There was no effect of rotation on 2-year-old leys when all years were considered.

N fixation was not significantly different in 3-year-old leys in 1999 and in 2000. The seasonal trends in N fixation were significantly different in 1999 and 2000 (REML p<0.001, Figure 4.5.7, Figure 4.5.8). There was also a significant effect of rotation on the seasonal trends. The 3-year-old leys in the 66% ley rotation had higher N fixation than the 3-year-old leys in the 50% ley rotation (REML p<0.001, Figure 4.5.9). This effect was more marked in 2000 than 1999 (REML p<0.001).



Figure .9 N fixation excluding transfer in 3-year-old leys in the 66% ley rotation and the 50% ley rotation in 2000. Values represent the accumulated N fixed during the month of sampling (mean of 6 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

Aldroughty

NfixET at Aldroughty was closely related to the dry matter yield of clover (Figure 4.5.3, Figure 4.5.4). In both years the rate of N fixation per unit of clover dry matter was low, because of low %NC and pNdfa, although in 1999 these were low at Tulloch also. Total N fixation in 1-year-old leys at Aldroughty over the sampling period was 48.8 kg N ha^-1 in 1999 and 49.34 kg N ha^-1 in 2000. N fixation was not significantly different in the two years. The estimated total annual N fixation, including N fixation below cutting height and outside the growing season (see above) was 107.148 kg N ha^-1 yr^-1 in 1999 and 109.04 kg N ha^-1 yr^-1 in 2000.

There were no significant differences between N fixation in 1-year-old leys at Aldroughty and Tulloch (Figure 4.5.7, Figure 4.5.8), and the seasonal patterns were similar (Figure 4.5.7, Figure 4.5.8). There was no significant effect of crop rotation on N fixation at Aldroughty.

2. Proportion of nitrogen derived from the atmosphere

Tulloch

When data from all years and ages were compared, pNdfa varied with year and/or age (REML p<0.001). pNdfa also showed a seasonal pattern (REML p<0.001), which varied with year and/or age (REML p<0.001).There were also significant differences between the two rotations in some years and/or ages of ley (REML p<0.005). To understand these patterns better, data from each year and age of ley were examined separately:

In 1997 pNdfaET showed a distinct seasonal pattern, declining in autumn (Figure 4.5.10, REML p<0.001). There was no overall effect of rotation on pNdfaET, but in August and September pNdfa in the 66% ley rotation was lower than pNdfa in the 50% ley rotation (REML p<0.01). Over the whole sampling period, pNdfa was on average 0.79 (calculated as total N fixation/total clover N yield). This is likely to be a low estimate because sampling started late in 1997.



Figure .10 Proportion of nitrogen in clover herbage derived from atmosphere, excluding transfer (pNdfaET) in 1-year-old leys at Tulloch in 1997. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1998 pNdfa followed a similar seasonal pattern to 1997, declining significantly in September and October (REML p<0.001, Figure 4.5.11). 1- and 2-year-old leys did not have significantly different values of pNdfa and there were no significant differences between the two rotations. Over the whole growing season, average pNdfa was 0.94 in 1-year-old leys and 0.95 in 2-year-old leys.



Figure .11 Proportion of nitrogen in clover herbage derived from atmosphere, excluding transfer (pNdfaET) in 1- and 2-year-old leys at Tulloch in 1998. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1999 pNdfaET followed a similar seasonal trend to previous years, declining in October (REML p<0.001, Figure 4.5.12). pNdfaET was highest in 1-year-old leys and lowest in 3-year-old leys (REML p<0.001). pNdfaET was higher in the 50% ley rotation, in the 1 and 3-year-old leys (REML p<0.01). Over the whole growing season, pNdfaET was 0.95, 0.95 and 0.92 in 1-, 2- and 3-year-old leys, respectively.



Figure .12 Proportion of nitrogen in clover herbage derived from atmosphere, excluding transfer (pNdfaET) in 1,2 and 3-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 1999. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 2000, pNdfaET followed a similar seasonal trend to previous years, declining significantly in autumn (REML p<0.001, Figure 4.5.13). When pNdfaET was compared in 1-, 2-, 3- and 4-year-old leys in the 66% ley rotation were compared, there was no significant difference between leys of different ages. Over the whole growing season, pNdfa was on average 0.91, 0.92, 0.85 and 0.9 in 1-, 2-, 3- and 4-year-old leys, respectively.



Figure .13 Proportion of nitrogen in clover herbage derived from atmosphere, excluding transfer (pNdfaET) in 1-, 2-, 3- and 4-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 2000. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

pNdfaET values in 1-year-old leys were compared in all years: 1997 and 2000 had significantly lower pNdfaET than other ages of ley (REML p<0.001, Figure 4.5.10, Figure 4.5.11, Figure 4.5.12 and Figure 4.5.13). There was significant seasonal variation (REML p<0.001), and this varied from year to year (REML p<0.001). In 1999, pNdfaET remained relatively high into October (Figure 4.5.12), unlike other years. There was also a significant year ´ month ´ rotation interaction (REML p=0.001), because of low pNdfa in the 50% ley rotation in September 1997.

When 2-year-old leys in 1998, 1999 and 2000 were compared, there was significant seasonal variation (REML p<0.001), which varied significantly between years (REML p<0.001), in 2000, pNdfaET declined unexpectedly in late June-early July (Figure 4.5.13). There was also a significant interaction between year, month and rotation (REML p<0.005), caused by slightly lower pNdfa values in the 66% ley rotation, in 1-year-old leys late in 2000, when compared to the 50% ley rotation.

pNdfaET in 3-year-old leys was lower in 2000 than in 1999 (REML p<0.001). The seasonal pattern also differed between the two years: in 2000 pNdfa declined steadily through the growing season, whereas in 1999 it remained high until October (REML p<0.001, Figure 4.5.12 and Figure 4.5.13). There was a slight difference between the seasonal patterns in the two rotations in 2000, pNdfa in the 66% ley rotation was lower than pNdfa in the 50% ley rotation late in the year (REML p<0.005).

Aldroughty

pNdfaET at Aldroughty was not significantly different from pNdfa at Tulloch in the years and ages of leys studied (Figure 4.5.12. and Figure 4.5.13). The seasonal patterns of pNdfa at both sites were also similar. pNdfaET was significantly higher in 1999 (average 0.96) than in 2000 (average 0.91, ANOVA p<0.01).

3. Nitrogen concentration in clover herbage

Tulloch

When data from all years and ages of ley was considered, %NC varied significantly with year and/or age (REML p<0.001). %NC followed a seasonal pattern (REML p<0.001), which varied with year and/or age of ley (REML p<0.001). There were also significant differences between the two rotations in some year or ages of ley (REML p=0.01). Data from each year and age of ley were re-analysed separately:

In 1997, percentage nitrogen in clover (%NC) varied significantly over the growing season, peaking in September and at a minimum in July (REML p<0.001). There was no effect of rotation on %NC in 1997. %NC was on average 4.35% (calculated as: total clover N yield / total clover dry matter). This was likely to be an overestimate of the average annual nitrogen concentration, because sampling began late in 1997.



Figure .14 Concentration of nitrogen in clover herbage expressed as a percentage of dry matter. 1-year-old leys at Tulloch, in 1997. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1998, %NC varied with season in a similar way to the previous year (REML p<0.001, Figure 4.5.15). There was no significant difference between 1- and 2-year-old leys overall, but 2-year-old leys had significantly higher %NC in June than 1-year-old leys (REML p<0.001, Figure 4.5.15). The average %NC, in 1998 at Tulloch was 3.7 in 1-year-old leys and 3.85 in 2-year-old leys.



Figure .15 Concentration of nitrogen in clover herbage expressed as a percentage of dry matter. 1- and 2-year-old leys at Tulloch, in 1998. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1999, %NC was significantly lower in 2-year-old leys than other ages of ley (REML p<0.001, Figure 4.5.16). There was significant seasonal variation (REML p<0.001) similar to that seen in 1997. The seasonal pattern in 3-year-old leys was slightly different from that of the other ages of ley, because %NC was relatively low at the start of the growing season and relatively high at the end of the growing season in 3-year-old leys (REML p<0.001, Figure 4.5.16). There was no effect of rotation on %NC in 1999. The average %NC values in 1999 at Tulloch were 3.1, 2.87 and 3.87 in 1-, 2- and 3-year-old leys, respectively.



Figure .16 Concentration of nitrogen in clover herbage expressed as a percentage of dry matter. 1-, 2- and 3-year-old leys at Tulloch and 1-year-old leys at Aldroughty, in 1999. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 2000, %NC was again significantly lower in 2-year-old leys (REML p<0.001, Figure 4.5.17). %NC showed a significant seasonal pattern (REML, p<0.001, Figure 4.5.17), similar to previous years. Three-year-old leys showed the greatest seasonal variation, and 1 and 4-year-old leys varied least over the course of the growing season (REML p<0.001, Figure 4.5.17). %NC was also higher in the 66% ley rotation in 1-year-old leys (REML p<0.001). Average %NC in 2000 at Tulloch was 4.34, 3.79, 4.38 and 4.21 in 1-, 2-, 3- and 4-year-old leys, respectively.



Figure .17 Concentration of nitrogen in clover herbage expressed as a percentage of dry matter. 1-, 2-, 3- and 4-year-old leys at Tulloch, and 1-year-old leys at Aldroughty in 2000. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

When only 1-year-old leys were examined, %NC was found to be lowest in 1999 and highest in 1997 and 2000 (REML p<0.001). The seasonal variation was greatest in 1998 (REML p<0.001). There was no significant effect of rotation on 1-year-old leys overall.

In 2-year-old leys, %NC was lower in 1999 (REML p<0.001). The seasonal pattern in 2-year-old leys varied between years, with most seasonal variation in 2000 and least in 1999 (REML p<0.001).

In 3-year-old leys, there was significantly higher %NC in 1999 than in 2000 (REML p<0.001). The seasonal pattern was also significantly different, between the two years: in 1999 %NC was at a minimum in July, whereas in 2000 %NC was high in July and fell to a minimum in August (REML, p<0.001, Figure 4.5.16 and Figure 4.5.17). %NC was higher in the 66% ley rotation than the 50% ley rotation in 3-year-old leys (REML p=0.01), especially early in the growing season (REML p<0.001) and in 1999 (REML p=0.01).

Aldroughty

The concentration of N in clover was significantly higher in 2000 than in 1999 at Aldroughty (ANOVA p>0.001, Figure 4.5.16 and Figure 4.5.17). %NC in 1-year-old leys at Aldroughty was not significantly different from %NC in 1-year-old leys at Tulloch. The seasonal patterns of the concentration of N in clover were also similar at Tulloch and Aldroughty in 1999 and 2000 (Figure 4.5.16 and Figure 4.5.17). Average N concentration at Aldroughty was 3.038% in 1999 and 4.15% in 2000

 

4. Dry matter yield of clover

Tulloch

When data from all years and ages of leys was considered, clover yield varied significantly with year and/or age (REML p<0.001) and season (REML p<0.001). The seasonal pattern of growth also varied with year and/or age (REML p<0.001). There was no overall effect of rotation on clover yield, but in some years and/or ages of ley there were differences between the two rotations (REML p<0.001). Data from each year and age of ley were re-analysed individually:

In 1997, the dry matter yield of clover showed a significant seasonal trend (REML p<0.01) which was very similar to the seasonal trend of N fixation (Figure 4.5.5 and Figure 4.5.18). There was no effect of rotation type on clover yield in 1997. Total clover yield in 1997 was 1211.3 kg dry

matter ha^-1. This is a low estimate because sampling began one month later in 1997 than in other years.



Figure .18 Clover herbage dry matter yield at Tulloch in 1-year-old leys in 1997. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1998, clover yield did not vary significantly between 1- and 2-year-old leys. There were similar seasonal trends to N fixation (Figure 4.5.6 and Figure 4.5.19), and these were significantly different in the two ages of ley (REML p<0.001), because the peak of growth in 2-year-old leys occurred later than the peak of growth in 1-year-old leys (Figure 4.5.19). There was no effect of rotation type on clover yield in 1998. Total yields of clover over the sampling period were 756.1 and 869.8 kg dry matter ha^-1 in 1- and 2-year-old leys, respectively. Over the whole year, the yields would have been higher, because of growth of clover before the start of the growing season. In 2-year-old leys, the total annual clover yield was 1120.6 kg dry matter

ha^-1. It was not possible to calculate this in the 1-year-old leys because of grazing prior to the start of the experiment.



Figure .19 Clover herbage dry matter yield at Tulloch in 1- and 2-year-old leys in 1998. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1999, clover yield was significantly higher in 2-year-old leys than in other ages of ley (REML p<0.001, Figure 4.5.20). There was a clear seasonal trend, with peak clover production in mid – late summer (Figure 4.5.20), closely following the seasonal pattern of N fixation. The seasonal patterns of clover growth were not the same in all ages of ley: yield in 2 and 3-year-old leys declined after midsummer, but yields in 1-year-old leys remained high into August (Figure 4.5.20). Total clover yields for the sampling period were 2020.5, 2809.7 and 1778.4 kg dry matter ha^-1 in 1-, 2- and 3-year-old leys, respectively. The total annual clover yield (including the initial sample, in ungrazed plots) was 3681.8 kg dry matter ha^-1 and 2491.7 kg dry matter ha^-1 for 2-year-old and 3-year-old uncut leys, respectively. There was no significant effect of rotation type on clover yield in 1999.



Figure .20 Clover herbage dry matter yield in 1-, 2- and 3-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 1999. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 2000, there was no significant difference between the clover yields in 1,2 and 3-year-old leys, but yields in 4-year-old leys were significantly lower (REML p=0.005). There were strong seasonal trends, similar to those of N fixation (REML p<0.001, Figure 4.5.8 and Figure 4.5.21). Clover yield in 3 and 4-year-old leys started to decline earlier in the year than clover yield in 1- and 2-year-old leys (REML p<0.001, Figure 4.5.21). Clover yields in 1 and 3-year-old leys were higher in the 66% ley rotation than the 50% ley rotation (REML p<0.01). Clover yields for the sampling period were 1909.4, 2032.5, 1878.4 and 985.9 kg dry matter ha^-1 for 1-, 2-, 3- and 4-year-old leys respectively. Total annual clover yields (including initial sample, ungrazed plots only) were 2830.8, 2089.1 and 1509.7 kg dry matter ha^-1 in 2-year-old, 3-year-old (ungrazed) and 4-year-old leys, respectively.



Figure .21 Clover herbage dry matter yield in 1-, 2-, 3- and 4-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 2000. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

When clover yields in 1-year-old leys were compared across all years, yields were lowest in 1998 and highest in 1999 (REML p<0.001). Seasonal variation of clover yield varied from year to year in a manner similar to N fixation (REML p<0.001, Figure 4.5.18, Figure 4.5.19, Figure 4.5.20 and Figure 4.5.21,).

In 2-year-old leys, clover yield was highest in 1999 and lowest in 1998 (REML p<0.001). Clover yield followed similar seasonal patterns to N fixation (Section 4.5.1). There was no effect of rotation on 2-year-old leys. Clover yield was not significantly different in 1999 and 2000 in 3-year-old leys. The seasonal patterns were generally similar to the seasonal patterns of N fixation. 3-year-old leys in the 66% ley rotation had higher clover yields than 3-year-old leys in the 50% ley rotation.

Data from Tulloch on ground cover of clover in Autumn, showed broadly similar patterns of age related variation to the patterns of clover yield observed in this study (Figure 4.5.22), 2-year-old leys generally had the highest clover ground cover. There was also considerable variation in clover ground cover between years.



Figure .22 Percentage ground cover of clover in autumn in successive years since the establishment of the trial. Each point is a mean of four samples (SAC data)

Aldroughty

Clover yields in 1-year-old leys at Aldroughty were not significantly different from clover yields in 1-year-old leys at Tulloch in 1999 or 2000. The clover yields in 1999 at Aldroughty were not significantly different from the clover yields in 2000 (Figure 4.5.20 and Figure 4.5.21). There was no significant effect of rotation on clover yields at Aldroughty. Total clover yields were 2809.1 kg dry matter ha^-1 in 1999 and 1445.9 kg dry matter ha^-1 in 2000.

5. Soil derived N in grass and clover

Tulloch

Overall, soil derived N (SDN) in grass and clover varied significantly between different years and/or ages of ley (REML p<0.001). There was a significant seasonal pattern (REML p<0.001), which varied between different years and/or ages of ley (REML p<0.001). The two rotations also had different levels of soil derived N overall (REML p<0.01), especially in certain years and/or ages of ley (REML p<0.001). To understand these patterns better, each year and age of ley were examined individually:

In 1997, Soil derived N declined over the course of the growing season (REML p<0.001, Figure 4.5.23). There was no significant effect of rotation on the leys studied in 1997. Soil derived N over the whole growing season in 1997, 1-year-old leys was 109.83 kg N ha^-1. The total annual SDN would have been higher than this, because sampling began late in 1997. 95.6% of SDN was accounted for by N in grass (grass N yield).



Figure .23 Soil derived N in grass and clover herbage in 1-year-old leys at Tulloch in 1997. Values represent the total accumulation of non-fixed N during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1998, soil derived N was similar in 1- and 2-year-old leys (Figure 4.5.24). There was a general decline in SDN over the course of the growing season (REML p<0.001, Figure 4.5.24), although in August, SDN was high in 2-year-old leys and low in 1-year-old leys (REML p<0.001, Figure 4.5.24). Total SDN was 59.4 kg N ha^-1 in 1-year-old leys and 69 kg N ha^-1 in 2-year-old leys. These are likely to be low estimates for total annual SDN, as there was considerable grass growth prior to the first sample. The total grass N yield (including the initial sample) in 2-year-old leys was 106.3 kg N ha^-1 (1-year-old leys had been grazed prior to the start of sampling, so early season grass yields couldn’t be measured). Grass N yield accounted for 94.1% and 97.6% of SDN in 1- and 2-year-old leys, respectively.



Figure .24 Soil derived N in grass and clover herbage in 1- and 2-year-old leys at Tulloch in 1998. Values represent the total accumulation of non-fixed N during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1999, SDN was highest in 3-year-old leys (REML p<0.01, Figure 4.5.25). SDN was 47.22, 46.3, and 65.19 kg N ha^-1 in 1-, 2- and 3-year-old leys respectively. Grass N yield accounted for 93.9% of SDN in 1-year-old leys 90.3% of SDN in 2-year-old leys and 95.1% of SDN in 3-year-old leys. These are low estimates for total annual SDN, as there was considerable grass growth early in the season. Total annual grass N yield (including the initial sample) was 72.7 kg N ha^-1 in 2-year-old leys and 89.1 kg N ha^-1 in 3-year-old cut leys. SDN was low at the beginning and end of the growing season (REML p<0.001). The largest seasonal variation was seen in the 3-year-old leys (REML p<0.001, Figure 4.5.25). There was no significant effect of rotation on SDN in 1999.



Figure .25 Soil derived N in grass and clover herbage in 1-, 2- and 3-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 1999. Values represent the total accumulation of non-fixed N during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 2000, SDN was lowest in 2-year-old leys and highest in 3-year-old leys (REML p<0.001, Figure 4.5.26). SDN was 111.8, 72.7, 139.0 and 114.0 kg N ha^-1 in 1-, 2-, 3- and 4-year-old leys, respectively. Grass N yield accounted for 92.6% of SDN in 1-year-old leys, 91.6% of SDN in 2-year-old leys, 90.8% of SDN in 3-year-old leys and 96.5% of SDN in 4-year-old leys. Total annual SDN was probably considerably higher than the figure given above. Total grass N yield (including initial sample) was 105.2 kg N ha^-1 in 2-year-old leys, 165.9 kg N ha^-1 in 3-year-old cut leys and 171.8 kg N ha^-1 in 4-year-old leys. SDN was highest in late summer (REML p<0.001, Figure 4.5.26), but the seasonal pattern varied greatly between different ages of ley. In 4-year-old leys SDN reached a maximum in late August-early September, approximately one month after younger leys (REML p<0.001, Figure 4.5.26).

SDN was higher in leys in the 66% ley rotation, than the 50% ley rotation (REML p<0.001) especially in 3-year-old leys (REML p=0.01, Figure 4.5.27).



Figure .26 Soil derived N in grass and clover herbage in 1-, 2-, 3- and 4-year-old leys at Tulloch and 1-year-old leys at Aldroughty in 2000. Values represent the total accumulation of non-fixed N during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

 

 

 

 



Figure .27 Soil derived N in grass and clover herbage (kg N ha^-1) in leys in the 66% ley rotation and the 50% ley rotation. Columns represent total soil derived N over the growing season (mean of 6 subplots). Bars represent standard errors.

In the 1-year-old leys, SDN was highest in 1997 and 2000 and lowest in 1998 and 1999 (REML p<0.001, Figure 4.5.23-Figure 4.5.26). There was no significant seasonal trend when all 1-year-old leys were considered, but there was a clear seasonal trend in 1997 (REML p<0.001, Figure 4.5.23).

There was no effect of rotation on SDN when only 1-year-old leys were considered.

In the 2-year-old leys, SDN was lowest in 1999 (REML p<0.001, Figure 4.5.25-Figure 4.5.27). SDN was low at the beginning and end of the growing season and peaked in late summer in 2-year-old leys (REML p<0.001, Figure 4.5.24-Figure 4.5.26), but in 2000 SDN rose slightly at the end of the year (REML p<0.001).

In the 3-year-old leys, SDN was significantly higher in 2000 than in 1999 (REML p<0.001, Figure 4.5.25 and Figure 4.5.26). SDN was low at the beginning and end of the growing season in both 1999 and 2000 (REML p<0.001) but in 1999, SDN was also low in late summer (REML p<0.001), Figure 4.5.25 and Figure 4.5.26). SDN was higher in the 66% ley rotation (REML p<0.001), especially early in the year (REML p<0.001). The difference was greatest early in the growing season in 2000 (REML p<0.005).

Aldroughty

SDN in 1-year-old leys was significantly higher in 2000 than 1999 at Aldroughty (ANOVA p<0.05, Figure 4.5.25 and Figure 4.5.26). There were no significant differences between SDN at Tulloch and Aldroughty in 1999. In 2000, SDN was significantly higher at Tulloch than Aldroughty (ANOVA p<0.001, Figure 4.5.26).

6. Nitrogen concentration of grass

Tulloch

When data from all years and ages of leys was examined, the nitrogen concentration of grass (%NG) varied significantly between years and/or ages of ley (REML p<0.001). There was a significant seasonal component to the variation (REML p<0.001) which varied with year and/or age of ley. %NG was also affected by the type of rotation (REML p<0.01), and this effect was stronger in some years and/or ages of ley than others (REML p<0.001). In order to clarify this, the data for each year and age of ley were examined separately.

In 1997, %NG was significantly higher in late summer and autumn than in early summer (REML p<0.001, Figure 4.5.28). There was no effect of rotation type on %NG in 1997. Average %NG (calculated as total grass Nyield / total grass dry matter yield) was 3.38%



Figure .28 Concentration of nitrogen in grass herbage expressed as a percentage of dry matter. 1-year-old leys at Tulloch, in 1997. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1998, average %NG in 1-year-old leys was 2.39% and in 2-year-old leys, 2.52%. These values were not significantly different. %NG rose over the course of the growing season (REML p<0.001, Figure 4.5.29). In the 2-year-old leys, %NG was relatively high in June, compared to the 1-year-old leys (REML p<0.001). There was no effect of rotation on %NG in 1998.



Figure .29 Concentration of nitrogen in grass herbage expressed as a percentage of dry matter. 1- and 2-year-old leys at Tulloch, in 1998. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1999, %NG was significantly higher in 3-year-old leys than in other ages of ley (REML p<0.005, Figure 4.5.30). %NG rose over the course of the growing season (REML p<0.001), and rose most sharply in the 3-year-old leys (REML p<0.01, Figure 4.5.30). %NG was significantly higher in the 66% ley rotation in 1999 (REML p<0.001). Average %NG for the growing season was 1.92, 1,9 and 2.01 in 1-, 2- and 3-year-old leys respectively.



Figure .30 Concentration of nitrogen in grass herbage expressed as a percentage of dry matter. 1-, 2- and 3-year-old leys at Tulloch, and 1-year-old leys at Aldroughty in 1999. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 2000, %NG was highest in 3-year-old leys and lowest in 2-year-old leys (REML p<0.001, Figure 4.5.31). %NG rose over the course of the growing season (REML p<0.001, Figure 4.5.31). This seasonal pattern was not the same for all ages of ley: 3-year-old leys had relatively high %NG at the start of the season, but at the end of the year had lower %NG than other ages of ley. 2-year-old leys had low %NG to begin with and high %NG at the end of the year (REML p<0.001, Figure 4.5.31). In 3-year-old leys, %NG was significantly higher in the 66% ley rotation than the 50% ley rotation (REML p<0.01, data in appendix). Average %NG for the whole growing season was 3.04, 2.59, 3.19 and 2.99 for 1-, 2-, 3- and 4-year-old leys, respectively.

 



Figure .31 Concentration of nitrogen in grass herbage expressed as a percentage of dry matter. 1-, 2-, 3- and 4-year-old leys at Tulloch, and 1-year-old leys at Aldroughty in 2000. Columns represent means of 12 subplots. The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

%NG in 1-year-old leys, was highest in 1997 and 2000 and lowest in 1999 (REML p<0.001, Figure 4.5.28-Figure 4.5.31). %NG rose as the growing season progressed (REML p<0.001) and this increase was most marked in 1998 (REML p<0.001, Figure 4.5.28-Figure 4.5.31). There was no effect of rotation on %NG overall, but %NG was higher in August 1997 in the 66% rotation than the 50% rotation (REML p=0.001).

In 2-year-old leys, %NG was significantly lower in 1999 (REML p<0.001, Figure 4.5.29 -Figure 4.5.31). %NG rose over the course of the growing season in 2-year-old leys (REML p<0.001), and rose most sharply in 2000 (REML p<0.001). There was no effect of rotation on 2-year-old leys.

In 3-year-old leys, %NG was significantly higher in 2000 than 1999 (REML p<0.001, Figure 4.5.30 and Figure 4.5.31). %NG rose significantly over the course of the growing season (REML p<0.001). In 1999, %NG was much lower at the start of the growing season than in 2000 (REML p<0.001). %NG was higher in the 66% ley rotation than the 50% ley rotation in 3-year-old leys (REML p<0.001), especially in 2000 (REML p<0.005).

Aldroughty

%NG was significantly higher in 2000 (2.66%) than 1999 (2.17%) at Aldroughty (ANOVA p<0.001, Figure 4.5.30 and Figure 4.5.31). The concentration of N in grass in 1-year-old leys at Aldroughty was higher than at Tulloch in 1999 (ANOVA p<0.05, Figure 4.5.30 and Figure 4.5.31) and lower in 2000 (ANOVA p<0.001, Figure 4.5.30 and Figure 4.5.31).

7. Dry matter yield of grass

When the entire data set was analysed, grass yield varied significantly between different years and/or ages of ley (REML p<0.001). Grass yield was low in Autumn (REML p<0.001), but this seasonal pattern varied significantly with year and/or age of ley (REML p<0.001). Grass yield was also affected by the type of rotation in some years and/or ages of ley (REML p=0.001). In order to understand these patterns more clearly, each year and age of ley was examined individually.

In 1997, grass yield declined significantly over the course of the growing season (REML p<0.001, Figure 4.5.32). Total grass yield for the sampling period was 3105.5 kg dry matter ha^-1. The total annual grass yield would have been much higher, because sampling began late in 1997. There was no effect of rotation on grass yield in 1997.



Figure .32 Grass herbage dry matter yield at Tulloch in 1-year-old leys in 1997. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1998, grass yield was 2328.2 and 2658.8 kg dry matter ha^-1 for 1- and 2-year old leys, respectively. These values were not significantly different. Grass dry matter yields declined as the growing season progressed (REML p<0.001). Grass yields in late summer were lower in 1-year-old leys than 2-year-old leys (REML p<0.001, Figure 4.5.33). There was no effect of rotation on grass yield in 1998.



Figure .33 Grass herbage dry matter yield at Tulloch in 1- and 2-year-old leys in 1998. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 1999, grass dry matter yields for the sampling period were 2308.2, 2190.4 and 3033 kg ha^-1 in 1-, 2- and 3-year-old leys, respectively. These values were not significantly different at p<0.01. Total annual grass yields (including initial sample, ungrazed plots only) were 4786.1 and 5878.5 kg dry matter ha^-1 for 2-year-old and 3-year-old (ungrazed) leys, respectively. Grass yields were at a maximum in July, and subsequently declined (REML p<0.001). The seasonal variation was greatest in 3-year-old leys (REML p<0.001, Figure 4.5.34). There was no effect of rotation on grass yields in 1999.



Figure .34 Grass herbage dry matter yield at Tulloch in 1-, 2- and 3-year-old leys at Tulloch, and 1-year-old leys at Aldroughty in 1999. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.

In 2000, grass yields during the sampling period were 3419.2, 2578.7, 3894.9 and 3693.1 kg dry matter ha^-1 in 1-, 2-, 3- and 4-year-old leys, respectively. Grass yields were significantly lower in 2-year-old leys (REML p<0.001, Figure 4.5.35). Grass yields were low at the beginning and end of the growing season (REML p<0.001, Figure 4.5.35). This seasonal pattern varied considerably with age of ley: grass yield in 3-year-old leys peaked in June, whereas grass yields in 4-year-old leys peaked in late August/ early September (REML p<0.001). Grass yield was significantly higher in the 66% ley rotation in 2000 (REML p<0.001, Figure 4.5.36).



Figure .35 Grass herbage dry matter yield at Tulloch in 1-, 2-, 3- and 4-year-old leys at Tulloch, and 1-year-old leys at Aldroughty in 2000. Values represent the herbage produced during the month of sampling (mean of 12 subplots). The period of measurement is the period from the start of regrowth on the earliest treatment to be harvested, to cutting of the last treatment to be harvested in each month. Bars represent standard errors.



Figure .36 Grass dry matter yields in the 66% and 50% ley rotations in 2000. Columns represent total grass dry matter yield for the sampling period (mean of 6 subplots). Bars represent standard errors.

In 1-year-old leys, grass yields were highest in 1997 and 2000 (REML p=0.001). Grass yields varied seasonally, declining in the autumn (REML p<0.001), and the seasonal pattern varied significantly: in 1997, yields were very high in July, and subsequently relatively low compared to other years (REML p<0.001). There was no effect of rotation on 1-year-old leys.

Grass yields in 2-year-old leys did not vary significantly between years, although the seasonal trends (REML p<0.001) did. Grass yield was lower at the start of sampling in 1999 than in 1998 or 2000 (REML p=0.001). There was no effect of rotation on grass yields in 2-year-old leys.

Grass yields in 3-year-old leys were significantly higher in 2000 than in 1999 (REML p=0.001). Grass yield was at a maximum in July (REML p<0.001), but in 1999 it declined more sharply in late summer/autumn than in 2000 (REML p<0.001). Grass yields were higher in the 66% ley rotation in 3-year-old leys (REML p<0.001), especially early in the growing season (REML p<0.001) and early in 2000 (REML p<0.001).

Aldroughty

Grass dry matter yields in 1-year-old leys at Aldroughty were 2102.1 kg ha^-1 in 1999 and 1705.8 kg ha^-1 in 2000. Dry matter yield of grass was significantly lower in 1-year-old leys at Aldroughty than Tulloch in 2000 (ANOVA p<0.001). Yield of grass was similar in both years at Aldroughty. Grass dry matter yields were similar in 1999 at both sites.

8. N transfer

Tulloch

N transfer calculated by isotope method a) was generally low or negative at Tulloch. In 2000 pNdfaIT was similar to pNdfaET in the 1,3 and 4-year-old leys, and lower in the 2-year-old leys (Table 4.5.1).

Method	Age of ley (years)	1997	1998	1999	2000
Isotope method a)	1	*6.30 (1.73)	1.10 (0.66)	0.78 (0.79)	1.15 (0.61)
	2		-0.25 0.18)	-5.04 (0.95)	*-3.13 (0.65)
	3			1.20 (0.65)	1.50 (2.11)
	4				0.21 (0.35)
Isotope method b)	1	0.04 (1.52)	1.31 (0.41)	1.84 (0.62)	26.7 (13.4)
	2		-4.17 (0.62)	-4.01 (1.49)	*-29.1 (3.43)
	3			1.18 (0.67)	-0.39 (2.92)
	4				11.87 (8.43)
N difference method	1	14.8(17.20)	-1.75 (9.24)	-13.5 (8.56)	28.12
	2		-36.0 (6.48)	-58.8 (5.08)	11.64
	3			9.69 (8.25)	19.87
	4				31.37

Table .1 N transfer at Tulloch (kg ha^-1) between 1997 and 2000, calculated by isotope methods a and b, and the N difference method. Numbers in brackets represent standard errors.

*July-October only

N transfer calculated by isotope method b) was also frequently low or negative (Table 22). In 2000, the 1-year-old ley had positive values, the 2-year-old ley had negative values, the 3 and 4-year-old leys had values not significantly different from zero.

N transfer calculated by the N difference method produced a lot of negative values between 1997 and 1999. In 2000, grass N yield in mixtures was greater than grass N yield in monocultures in all ages of ley (Table 22). N transfered in 2000 was approximately 35% of total above ground fixed N in 1-year-old leys, 13% in 2-year-old leys, 21% in 3-year-old leys and 50% in 4-year-old leys. Excluding non-fixed N in clover, N transfer of fixed N for 2000 was 30.72, 4.96, -5.27 and 46.76 % for 1-,2-,3- and 4-year-old leys, respectively. The low and negative N transfer values in 2 and 3-year-old leys, may indicate that in these stages of the rotation, clover is successfully competing with grass for soil N. N fixation above ground calculated by N difference for these plots in 2000 gave values of 62.72, 87.28, 83.15 and 62.82 kg ha^-1 for 1-, 2-, 3- and 4-year-old leys, respectively.

There was generally little agreement between the various methods for calculating N transfer. Table 4.3.1 shows the correlations between the various estimates of N transfer for 2000, when the N transfer measurements were felt to have been made with the greatest accuracy.

	Isotope method a	Isotope method b
Isotope method b	0.688
N difference method	0.104	0.398

Table .2 Correlations between N transfer estimates at Tulloch, calculated by isotope methods a) and b) and the N difference method. Data from plots 13, 16, 17 and 18 in 2000.

Aldroughty

In both 1999 and 2000, N transfer values calculated by isotope method a) were not significantly different from 0. N transfer calculated by isotope method b) was slightly negative in both years, and N transfer calculated by the N difference method was not significantly different from 0 in either year.

Method	1999	2000
Isotope method a)	-1.40 (0.59)	-4.36 (6.21)
Isotope method b)	-2.76 (1.06)	-3.05 (1.12)
N difference method	-0.45 (6.39)	-1.75 (13.90)

Table .3 N transfer calculated by isotope methods a) and b) and the N difference method at Aldroughty (kg ha^-1). Numbers in brackets represent standard errors.

9. Plant physiology

Flowerheads

Figure 4.5.37 shows the number of clover flowerheads per g of clover dry matter on all sampling dates in 2001. Total numbers of flowers and flowering rate averaged over the whole year are shown in Table 4.5.4.



Figure .37 Number of flowers per g dried clover in samples taken for ¹⁵N analysis in 2001. Columns represent the mean of 12 subplots. Bars represent standard errors.

Flowering rate was generally highest in 1-year-old leys (ANOVA p<0.001), and in July (ANOVA p<0.001, Table 4.3.1). Flowering was over by the end of September (Figure 4.5.37). There was no direct effect of rotation on the flowering rate of clover but, in 3-year-old leys, flowering rate was higher in the 66% ley rotation (ANOVA p<0.001). The flowering rate was also around 10% higher in the 66% ley rotation in July (p<0.05).

Age of ley (years)	Flowers/g clover dm
1	2.0 (0.27)
2	1.26 (0.21)
3	1.51 (0.45)
4	0.62 (0.23)

Table .4 Number of clover flowers produced per m² within the subplots used for the ¹⁵N dilution study. Numbers in brackets represent standard errors.

10. Tap-roots

In August 2001, only clover in 1-year-old leys had significant numbers of tap-roots. No tap-roots were observed in clover from 4-year-old leys, and tap-roots were rare in 2 and 3-year-old leys (Figure 13). These results should be taken with caution, as tap-roots were only counted in soil samples from one sampling date. Also, the results do not show what proportion of clover plants had tap-roots, but rather, the average number of tap-roots in 8 soil cores taken from clover microsites.

 



Figure .38 Dry matter content of clover herbage in 2001, expressed as a percentage of fresh weight. Columns represent average dry matter content for the whole year (mean of 4 subplots).

Other differences were also observed between different ages of ley. In 2001, clover in 1-year-old leys was observed to have a higher dry matter content than clover in 2, 3 and 4-year old leys (Figure 4.5.38).



Figure .39 Number of tap-roots present in 8 soil cores from clover microsites in August 2001. Columns represent the mean of four samples. Bars represent standard errors.

 

11. Nitrogen budgets

Year	Age of ley	N in grass in mixture (kg ha-1)	N in grass monoculture (kg ha-1)	Fixed N in clover (kg ha-1)	Soil N in clover (kg ha-1)	Total N in herbage (kg ha-1)
1997	1	121.5	121.8	41.8	11.1	151.7
1998	1	54.3	57.7	26.2	1.6	82.2
	2	67.4	105.1	30.3	1.7	99.4
1999	1	44.3	93.0	61.9	2.9	109.1
	2	41.8	103.6	75.6	4.5	121.9
	3	61.2	55.5	47.6	4.0	112.8
2000	1	103.5	86.6	73.7	7.9	185.1
	2	66.5	72.8	70.4	6.1	143.1
	3	126.1	155.0	69.2	12.8	208.2
	4	110.1	86.4	33.5	5.7	149.3
	total	360.2	368.8	153.2	22.5	513.2

Table .5 N budgets in all years and ages of ley, showing fixed and soil derived N present in all grass and clover mixtures and monocultures. Data from all plots.

Age of ley	N in grass mixture (kg ha-1)	N in grass monoculture (kg ha-1)	Fixed N in clover (kg ha-1)	Soil N in clover (kg ha-1)	N benefit to grass from clover (kg ha-1)
1	110.1	86.6	53.1	4.6	23.5
2	76.8	72.8	75.6	7.7	4.0
3	147.0	155.0	76.1	15.1	-8
4	114.1	86.4	31.4	3.8	27.7
total	448	400.8	236.2	31.2	47.2

Table .6 N budget for plots 13 (4-year-old ley), 16 (1-year-old ley), 17 (2-year-old ley) and 18 (3-year-old ley) showing fixed and soil derived N present in all grass and clover mixtures and monocultures.

N budgets for the subplots are shown in Table 4.5.5 and Table 4.5.6. Table 4.5.6 shows the N budget for selected subplots in 2000. lt was difficult to calculate N transfer in this experiment with any confidence. The N transfer calculations at Tulloch in 2000 were an attempt to improve the accuracy of the method, by reducing the distance between the grass-clover subplots, and the grass monocultures. For this reason, this section will mainly focus on the data from Tulloch in 2000.

The data for 2000 shows that harvesting of the subplots in the 1-,3- and 4-year old leys removed more soil N from the system than was fixed by clover, and considerably more than was apparently transferred from clover to grass. Grass yields are a low estimate, as there was considerable early season grass growth, equal to around 50% of the grass yield during the sampling period. For this reason, grass yields will be multiplied by 1.5 and clover yields by 1.3 in the following calculations: In the cut leys the net removal of soil N in grass and clover over the growing season (taking into account N fixation below ground) was 49.4, 181.9 and 110.7 kg N ha^-1 for 2, 3 and 4-year-old leys, respectively. Transfer of fixed nitrogen from clover calculated as N difference was 4, 0 and 27.7 kg N ha^-1 for 2, 3 and 4-year-old leys, respectively. These represent conditions under the exclusion cages. In the field, the effects of grazing and FYM applications must also be taken into account:

Applications of manure to cut plots in 2000 were 23, 10 and 15 tonnes ha^-1 in 2-, 3-, and 4-year-old cut leys, respectively. Assuming that composted organic FYM contains 25% dry matter and 5.9 % N (Watson, 2003), this means that 339.25, 147.5 and 221.25 kg N were applied as manure per hectare in 2-, 3- and 4-year-old cut leys, respectively. According to Granstedt (1995), approximately 58% of this will be lost to the atmosphere, making the manure inputs 142.5, 61.95 and 92.9 kg N ha^-1 in 2-, 3- and 4-year-old cut leys, respectively. This makes the approximate balance of N (inputs - output) 97.1, -127.9 and 9.94 kg N ha^-1 in 2, 3 and 4-year-old cut leys, respectively (including N fixation below ground, estimated at 70% of fixed N in herbage). This calculation does not take leaching into account.

In grazed leys around 25% of the nitrogen removed by grazing animals is likely to be retained by the animals, the rest returned as excreta (Petersen et al., 1956), although 58% of this excreta N is likely to be lost by volatilisation and leaching (Petersen et al., 1956; Granstedt, 1995). This would amount to a net removal of 59.3 and 106.18 kg N ha^-1 of soil derived N from grass and 3.14 and 10.3 kg N ha^-1 of soil derived N from clover in 1- and 3-year-old leys, respectively. Assuming that 75% of the fixed N in the clover herbage will be returned as dung and urine by the grazers, and 58% of this will be lost to air and water, this means that 40.2 and 39.56 kg fixed N ha^-1 will be returned to the soil in 1 and 3-year-old leys, respectively (including N transferred from clover to grass, in 1-year-old leys). In addition, N fixed in roots, stolons and stubble will be equal to approximately 70% of the fixed N in leaves and shoots of clover (Jorgensen and Ledgard, 1997). This will add an extra 37.16 and 53.3 kg N ha^-1 to 1- and 3-year-old leys respectively. This gives a total N balance of 14.92 and –33.9 kg N ha^-1 for 1- and 3-year-old leys, respectively. Ledgard (1991) estimated that in a sward grazed by dairy cattle in New Zealand, 22% of fixed N was transferred above ground from clover to grass. In the same study, below-ground transfer was estimated at 70 kg N ha^-1 (26% of fixed N).

This method can also be used to estimate the N budget for plots 3, 10, 13 and 19 over the 4 years of the study, as these plots were 1-year-old in 1997, 2-year-old in 1998 and 3-year-old in 1999. Plots 3 and 13 were also 4-year-old in 2000. Manure applications to 2-year-old leys in 1998 were 30 t ha^-1. Three-year-old cut leys in 1999 received 23 tonnes manure per ha and 4-year-old leys received 15 tonnes manure/ha. This means that the 50% ley rotation accumulated 433.8 kg N ha^-1 between 1997 and 1999, and the 66% ley rotation accumulated 370.2 kg N ha^-1 between 1997 and 2000. The 66% ley rotation accumulated less in the ley, but this was because of the high amount of grass removed in the 4-year-old leys in 2000, and the manure applied to the 50% ley rotation 3-year-old plots in 1999. Much of the N in grass and clover removed from these two rotations would subsequently be returned to the plots in the arable phase in the form of manure. First year oats at Tulloch received no N, 2nd year oats received 12 tonnes, and swedes received 12 tonnes. This means that 177 kg N ha^-1 are returned to the 66% ley rotation and 354 kg N ha^-1 are returned to the 50% ley rotation in the form of manure. 3.5-3.9

These estimates do not take into account grazing and decay of clover over winter. The evidence from the grass N yields (Table 4.5.6) suggests that N is accumulating over the first three years of the ley.

4. Discussion

1. Annual variation

Growth of grass and clover and the ratio of grass to clover are likely to be affected by the amount of sunlight available in any particular year. The competition between grass and clover for light is likely to be affected by the sward height, as ryegrass grows taller than clover when permitted to (Harris, 1987). Ryegrass grows at slightly lower temperatures than clover (Chestnutt and Lowe, 1970), and clover in Britain needs temperatures of 9^o C. to fix nitrogen (Munro, 1970). The result of this is that clover tends to start growing later in the season than ryegrass (Ollerenshaw and Baker, 1981), and early season competition between grass and clover is likely to be affected by availability of soil nitrogen as well as temperature. Soil moisture also affects growth of grass and clover. Clover is shallower rooting than ryegrass and therefore more prone to drought (Harris, 1987, Caradus, 1990). It would be expected that variations in rainfall between seasons would affect the growth of grass and clover (e.g. Schils et al., 1999), but a comparison of the N fixation and soil N data with the rainfall and temperature data for the site does not reveal any obvious patterns. 1998 and 2000 were wetter years than 1997 and 1999, and 2000 was colder than the other years studied (Figure 3.1.2-Figure 3.1.5). Neither rainfall or temperature were simply related to annual variation of N fixation, clover yield, clover N concentration, pNdfa, soil derived N in herbage, grass yield or grass N concentration.

2. Within-season variation

Clover N concentration and pNdfa followed clear seasonal patterns that were remarkably consistent in different sites, ages of ley, and years. pNdfa tended to decline late in the growing season, and a similar pattern was observed by Boller and Nosberger (1987). N concentration of clover was lowest in July and highest in September. In 2001, flowering rate was also highest in July, except in 4-year-old leys, which peaked in June. Assuming flowering rate followed a similar pattern in previous years, it would be negatively correlated with N concentration during the growing season. If intact white clover flowerheads prior to seed formation have a lower N concentration than leaves and petioles, this could explain the seasonal variation of clover N concentration and the variations of this pattern between different ages of ley. Warembourg et al. (1997) observed that flowers of red clover had a C:N ratio of approximately 15, lower than any other part of the plant, save nodules. However, the C:N ratio of flowering stems varied enormously from 10 to 70, and 100 when dead. Hogh-Jensen et al. (2001) observed that N concentration in clover was higher in petioles than in any other plant part. It is therefore possible that flowering might cause an overall decline in white clover N concentration.

Regular cutting increases clover N concentration (Elgersma et al. 2000), probably by suppressing flowering and stimulating new growth, so the N concentration values and flowering rates may not actually reflect the real situation in the field, especially in cut leys. If regular cutting increases clover N concentration, then it is likely that grazing might have the same effect. This could explain the elevated clover N concentration early in the season in 3-year-old leys in the 66% ley rotation, as these leys were grazed prior to the start of sampling.

Seasonal variation of grass N concentration may have been due in part to the flowering cycles of the grasses. Generally grass N concentration rose over the course of the growing season. Similar trends were observed by Elgersma et al. (2000). Flowering of grasses was not measured in this experiment.

3. Age of ley related variation

N fixation and pNdfa followed roughly opposite patterns to grass yield with respect to age of ley. N fixation was generally highest in 2-year-old leys, grass yield, soil derived N and grass N concentration were lowest in 2-year-old leys. pNdfa was lowest in 3-year-old leys, and grass yield was highest in 3-year-old leys. However, there was no clear inverse correlation between these factors. Dry matter of clover followed a similar pattern to N fixation (Figure 4.5.5-Figure 4.5.8, Figure 4.5.18-Figure 4.5.21).

It was hypothesised that the pattern of clover N concentration in different ages of ley could be explained by differences in the rate of flowering of clover with age of ley, possibly related to the life cycle of clover. When a clover plant makes the transition from a tap-rooted rosette to a creeping stoloniferous form, it would be expected that there would be changes in the proportion of stolon, leaf, flower and petiole that could cause changes in the N concentration of the sampled herbage (Warembourg et al., 1997). Clover in leys of different ages did show different rates of flowering. It should be stressed that this was only apparent in the controlled conditions under an exclosure cage. In the field, it was obvious that there was far more clover in flower in cut leys than in grazed leys, as sheep grazing consumed a large number of flowerheads. Flowering rate tended to decline with age of ley, and so did not appear to correlate with the clover N concentrations observed in different ages of ley in 2000, as clover flowers are rich in N (Warembourg et al., 1997). Thomas (1987) observed that clover plants fed entirely on mineral N produced more flowers than those fixing N, when grown with 16h of daylight in every 24 hours (Thomas, 1987).

Leys of different ages had different pNdfa values, but this had little effect on the N fixation values, which closely followed the pattern of clover yield (Figure 4.5.1-Figure 4.5.4, Figure 4.5.5-Figure 4.5.8, Figure 4.5.18-Figure 4.5.21). Grass N concentration followed a similar pattern to clover N concentration in leys of different ages, which could mean that both clover and grass are responding in a similar way to soil conditions in leys of different ages. From the limited data available, it seems that clover in this system loses its taproots between the first and second years of ley (when the clover plants are between two and three years old). The senescence of taproots immediately precedes the peak of clover yield, and a sharp reduction in clover N concentration. Frequency of taproots appears to follow a similar pattern to flowering of clover, both declining with age of ley in August. Pederson (1989) observed that clover plants grown with additional N had thicker taproots, more, longer and thicker stolons and more flowerheads than those without additional N. Ryle et al. (1981b) also observed enhanced growth of clover plants with added N, although this was in an experiment in which clover was grown as a monoculture, so the effects of elevated soil N and competition on clover morphology could not be observed. Neither of these experiments ran for long enough to observe loss of taproots, and the clover plants were not grown with grass, so the response of the clover to intense competition from grass in an N rich soil could not be measured. It would be expected that with additional nutrients and no competition, clover would show a general increase in size. It would be interesting to repeat the experiment using a grass clover mix, and a variety of N treatments, over a 4-year period. The disappearance of taproots could be observed either with a minirhizotron, or using destructive sampling. Stolon mass, flowering rate and/or seed yield of the clover, could also be measured, to test the hypothesis that clover alters its reproductive strategy in response to soil N and competition.

Data on ground cover of various species at Tulloch (Figure 4.5.22) broadly supports the evidence of an age-related pattern observed in this study. Overall, 2-year-old leys have most clover. However when the pattern is viewed over all years since conversion, it becomes clear that there have been considerable fluctuations over time. This is to be expected, as the plots were going through their first six-year rotation cycle since conversion while these measurements were being taken. For some reason, 1-year-old leys are less subject to these fluctuations than other ages of ley. This may be because the clover content in 1-year-old leys is determined more by the seeding rate than by any ecological factors. High clover seeding rates can improve white clover establishment, but the effect is overridden after a couple of years (Frame and Newbould, 1986).

4. Effects of cutting and grazing on clover and grass

It was not possible to directly measure the effect of cutting and grazing on nitrogen fixation, clover yield, N concentration of clover, pNdfa, soil derived N in herbage, grass yield or grass N concentration in this study. These variables were only measured in the experimental subplots which were all cut once every four weeks, and so unaffected by the cutting and grazing treatments of the surrounding plots. However, certain rotation effects were only detected mainly in 3-year-old leys (Age of ley x Rotation interactions), and were most prominent early in the year. It is possible that these result from grazing of the 3-year-old leys in the 66% rotation before sampling began. It is reasonable to suggest that in this system, grazing increases soil derived N, increases grass N concentration and dry matter yield, and reduces pNdfa. All of these effects could be caused by the above ground N transfer of N from clover to grass, and elevated soil N. Alternatively these effects could be due to the type of rotation. Schils et al. (1999) observed less white clover ground cover in a management system using rotational grazing and cutting than one with cutting only.

5. Effects of rotation type

Across all years and ages of ley, there was evidence that N concentration of grass was elevated in the 66% ley rotation. There was also evidence of elevated N fixation, %NC, SDN and %NG, and reduced pNdfa in the 66% ley rotation, mainly in 1 and 3-year-old leys, in some years. There was no difference between the two rotations in their clover N concentrations. Other evidence from the site showed that there was slightly more K in the 66% ley rotation and this effect was most pronounced in 1999 and 2000 (data in appendix).

Differences between the two rotations tended to be small. N fixation provided by the extra year of ley in the 66% ley rotation was not great, although the extra grass yield could be seen as a benefit to be offset against the loss of 1 years arable crop from the rotation.

6. Variation between sites

N fixET, pNdfa, grass N yield, clover dry matter and average clover N concentration were similar at Tulloch and Aldroughty. There were significant differences between grass N concentration and yield at Aldroughty and Tulloch. This may reflect differences in soil fertility or rainfall between the two sites. Between 1997 and 1999, Aldroughty had significantly lower P (REML p<0.001) and K (REML p<0.001) than Tulloch, and significantly higher magnesium (REML p<0.001, data in appendix). A deficiency of any of these nutrients could reduce N fixation. Legumes are known to have a higher demand for P & K than non-legumes, so a deficiency of these nutrients would be expected to reduce the clover:grass ratio The low grass N concentration may have been due to the relatively high proportion of broad-leafed weed species (Rumex spp., thistle, volunteer potato etc.) included in the grass fraction at Aldroughty, compared to Tulloch. The flowering rate of clover was not measured at Aldroughty, so a comparison could not be made.

7. Variation within sites

N fixation tended to be higher in block 1 at Tulloch than in block 2 (Data in appendix). This may be because block 2 was slightly higher than block 1, more sloping and drier. Ledgard et al. (1987) found lower N fixation by white clover on sloping sites, although this may have been due to competition from other legume species, which were not at Tulloch. The increased fixation manifested as an increase in clover yield, N concentration and pNdfa were similar in the two blocks. There were no differences in flowering rate of white clover between different blocks at Tulloch. Nitrogen concentration of clover and grass was not significantly different between blocks.

8. N transfer

Isotope method a) only considers differences between pNdfaIT and pNdfaET, and not differences in N yield of grass mixtures and monocultures. It is rarely used in the literature. In this experiment, differences between pNdfaIT and pNdfaET were negligible. Method a) compares isotopic enrichment in clover and grass in mixtures and monocultures, but does not consider differences in N yield of grass in mixtures and monocultures. Isotope method b) produced high estimates of N transfer for 1 and 4-year-old leys (36 and 47% of total N fixed, respectively), but 2-year-old leys had negative N transfer values, and 3-year-old leys had non-significant N transfer, calculated by this method. The N difference method also estimated N transfer to be highest in 1- and 4-year-old leys (35 and 50% of total N fixed respectively), but gave large estimates for the 2 and 3-year-old leys when non-fixed N in clover was taken into account (13 and 21 %, respectively). There is therefore some agreement between isotope method b) and the N difference method. The N difference method measures difference in nitrogen yield between grass in mixtures and monocultures, but this does not prove that any additional nitrogen in the grass in mixtures actually originated in the clover. Grass grown in mixtures often had a similar or even lower dry matter yield than grass in monocultures, but a higher N concentration. The effect of clover on grass is probably not a simple one. Discrepancies between the isotope and N difference methods could indicate that clover benefits ryegrass in other ways besides increasing nitrogen availability. Nitrogen released from clover roots might increase the turnover of soil organic matter and release soil nitrogen for grass to use. Any transfer of nitrogen from grass to clover might also complicate the picture, although there is little evidence for this happening in the literature.

1. Soil chemistry of grass-clover swards
1. Aims

The cyclical replacement hypothesis predicts that as the age of ley increases, soil N should increase and clover should retreat to a diminishing number of low N microsites (the term "microsites" is used here to mean patches within the sward at the scale of individual grass and clover plants). Soil N is also predicted to increase over the course of each growing season. Depletion of nutrients such as Phosphorus (P), Potassium (K) and Magnesium (Mg) would potentially have the same effect as elevated N: giving grass a competitive advantage over clover. It would be expected that the 66% ley rotation would have higher levels of readily available soil N because this rotation has a longer fertility building ley phase and correspondingly shorter arable phase.

Soil was sampled and analysed to compare concentrations of available soil N levels over time, in different ages of ley, under different management regimes, and in different rotations, to see if any of the variations in grass and clover behaviour could be attributed to variation of soil nitrogen. Phosphorus, potassium and magnesium levels were also examined, to a more limited extent for a similar purpose.

Soil N under grass and clover microsites was also analysed and compared over time, to test the hypothesis that clover colonises microsites which are low in soil N, and elevates soil N locally, before being out-competed by grass. As clover is sensitive to the combination of low temperatures and high soil N, it can be predicted that clover will die off back from microsites with high levels of available soil N during the colder months. Root and nodule turnover is likely to be highest when clover is actively growing, so it would be expected that any elevation of soil N in microsites occupied by clover would be most obvious during the summer.

2. Hypotheses

1. Soluble soil nitrogen increases with increasing age of ley, as a result of accumulation of fixed N transferred from clover to soil by above and below ground N transfer and decay of clover leaves petioles and stolons over winter.
2. The 66% ley rotation will have more available N than the 50% ley rotation, because of its higher ratio of fertility building ley to arable.
3. Clover containing microsites will have higher levels of available soil N than grass only microsites in summer, because of N transfer and root and nodule turnover during peak growth.
4. Clover containing microsites will have lower levels of available soil N than grass only microsites in winter, because clover is sensitive to low temperatures and high soil N and is expected to die back during winter.

2. Methods
1. Soil sampling and analysis

Soil was sampled approximately every 2 months from the trial rotation ley plots at Tulloch (Section 3.1). Plots were not sampled in the autumn immediately after harvest of the oats. Samples were taken on a grid basis, from the top 150 mm of soil. From each plot, 8 soil cores were taken from microsites containing only grass and weeds, and 8 cores were taken from microsites containing clover. Grass cores and clover cores were bulked to provide one grass soil sample and one clover soil sample for each plot. Cores were crumbled and any plants and roots were removed.

A sample of fresh soil from each sample (30-35g) was weighed, dried at

105 ^oC, and weighed again in order to measure the dry matter content of the soil. 10g of each bulked soil sample was accurately weighed into a screw topped plastic bottle and mixed with 50 ml of 0.5 m K₂SO₄. The bottles were then placed in a mechanical end-over-end shaker for 2 hours. The soil solutions were then filtered through a fluted filter paper (Whatman No. 42). and analysed for NH₄^-N and NO₃^-N as described below. A 50 ml sample of K₂SO₄ was also included as a blank. A 5 ml aliquot of each filtered extract was accurately pipetted into an autoclaveable universal bottle. 1 ml of persulphate oxidiser, made from 1.34g K₂S₂O₈ + 0.3g NaOH in 100 mls of water, was added to each aliquot, and the mixture was autoclaved at 110^oC for 30 minutes (Cabrera and Beare, 1993). The processed extract was then analysed for NO₃^-N at a dilution of 1.5 ml of extract to 0.09 ml of 0.1M NaOH as described below.

The extracts and the digested aliquots were accurately pipetted into 2 ml autoanalyser cups and analysed for nitrate-N and ammonium N using a Carlo Erba continuous flow autoanalyser (CE Instruments, Milan, Italy). The autoanalyser converts nitrate to nitrite in the presence of copper and cadmium and then measures the nitrite calorimetrically, at 540 nm by a diazotization reaction. Ammonia is measured by reacting ammonia with hypochlorite ions to form monochloramine, which is reacted with sodium salicylate in the presence of sodium nitroprusside to form an emerald-green, indo-phenol type compound which is measured calorimetrically at 660 nm. The autoanalyser produces measures of nitrate and ammonia as ppm N in solution.

The nitrate-N concentration of the digested samples gave a value for total soluble N (TSN). Soluble organic nitrogen (SON) was calculated from TSN –(ammonium + nitrate-N). Soil samples from 17^th August 2001, were also analysed for total extractable P, K and Mg. Samples were analysed immediately whenever possible, or frozen for analysis at a later date.

Standards for nitrate were prepared as follows: 0.9025g Potassium nitrate (dried at 98^oC for 1 hour), was dissolved in 1000 ml of de-ionised, distilled water, to produce a 125 ppm stock solution. This stock solution was then diluted to produce standards ranging from 0.1 to 2 ppm NO₃^--N.

Standards for ammonium were prepared as follows: 0.4722g of ammonium sulphate (dried overnight at 98^oC was dissolved in 1000 ml deionised water, to give a 100 ppm stock solution. This stock solution was diluted to produce standards ranging from 0.1 to 2 ppm NH₄⁺-N.

2. Calculations

Weight of oven dried soil = (weight of container + oven dried soil) – weight of container. [.1]

Weight of moisture = (weight of container + fresh soil) – (weight of container + oven dried soil) [.2]

Dry weight content = ((weight of container + dry soil)-weight of container) / ((weight of container + fresh soil) – weight of container) [.3]

Weight of oven dried soil/g = fresh sample weight/(1+(% moisture/100)) [.4]

where the fresh sample weight is 10g

In order to convert the output of the autoanalyser (ppm per sample) into kg N ha^-1 in the top 150 mm of soil, the following calculation was made:

[.5]

Where the dilution factor is 50, the fresh sample weight is 10g and the sample depth was 15 cm

Bulk density was calculated as:

Bulk density = soil dry wt/volume of fresh soil [.6]

3. Sampling dates

Sampling was performed on 28th January 1999, 8th April 1999, 21st June 1999, 16th August 1999, 12th October 1999, 17th January 2000, 28th February 2000, 6th April 2000, 4th July 2000, 27th September 2000, 29th November 2000, 4th April 2001, 18th June 2001 and 17th August 2001.

4. Statistical analysis

Nitrate-N, ammonium-N, SON and TSN data from all sampling dates was transformed and analysed by REML. Model terms were: age of ley, management type (cutting or grazing), season (month when sample was taken), rotation (50% or 66%), vegetation (grass or clover microsite) and year (1999, 2000 or 2001). Soluble organic N and ammonium data from June 1999 was omitted as the high ammonium values on this date (10-20 kg N ha^-1, compared to c.5 kg N ha^-1 in April and August 1999) suggested contamination of the K₂SO₄ solution. Samples where SON was negative (i.e. TSN values were inconsistent with nitrate-N and ammonium values) were also omitted. Nitrate-N values were rooted to the fourth power and ammonium and TSN were transformed by Log Fisher, to normalise the data.

3. Results
1. Soil Nitrate-N

Nitrate-N was generally higher under clover patches than under grass patches (REML p<0.01). This effect was far more pronounced in summer than in winter (REML p<0.001, Figure 5.4.1). Nitrate-N was approximately twice as high in summer 2000 as in summer 1999 (Figure 5.4.1, REML p<0.001).



Figure .1 Nitrate-N in soil under grass and clover microsites in all ley plots at Tulloch, between February 1999 and August 2001. Points represent means of 14 samples. Bars represent standard errors.

Nitrate-N was significantly lower in 1999 than in 2000 (REML p<0.001), especially during the growing season (REML p<0.001). Nitrate-N was 3-4 times higher during the winter than in the growing season (REML p<0.001, Figure 5.4.1 and Figure 5.4.2).

Nitrate-N was highest in 3-year-old leys and lowest in 2-year-old leys. (p<0.001). During the growing season, 3-year-old leys had approximately twice as much nitrate-N as 2-year-old-leys, but the difference was smaller early in the year (REML p<0.001). The age related pattern varied from year to year: nitrate-N in 1 and 4-year-old leys was relatively high compared to other ages of ley in 2000 and 2001, and relatively low compared to other ages of ley in 1999, (REML p<0.001, Figure 5.4.2).

During early summer, grazed leys had approximately 50% higher nitrate-N than cut leys (REML p<0.001, data in appendix). Nitrate-N levels were similar in both rotations.



Figure .2 Nitrate-N in soil in 1-, 2-, 3- and 4-year-old leys at Tulloch between January 1999 and August 2001. Columns represent means of 8 samples (1-, 2- and 3-year-old leys) and 4 samples (4-year-old leys). Bars represent standard errors.

2. Soil ammonium-N

Soil ammonium-N levels were highest in 2000 and lowest in 1999 (REML p<0.001, Figure 5.4.3), and varied from month to month, being 2-3 times as high at the end of the growing season (REML p<0.001), although this effect was less pronounced in 1999 (REML p<0.001, Figure 5.4.3).

Ammonium-N was highest in 3 and 4-year-old leys, approximately 50% higher than in 1 and 2-year-old leys (REML p<0.001, Figure 5.4.3). Ammonium was not significantly different in cut and grazed leys. Management (cutting or grazing) did not significantly affect ammonium. Clover and grass patches did not have significantly different ammonium levels. The 66% ley rotation had up to 50% higher ammonium than the 50% ley rotation (REML p<0.001, Figure 5.4.4).



Figure .3 Ammonium-N in soil in 1-, 2-, 3- and 4-year-old leys at Tulloch between January 1999 and August 2001. Columns represent means of 8 samples (1-, 2- and 3-year-old leys) and 4 samples (4-year-old leys). Bars represent standard errors.



Figure .4 Ammonium-N in the 50% and 66% ley rotations at Tulloch (1–3-year-old leys only). Columns represent means of values for 12 plots. Bars represent standard errors.

3. Soluble Organic Nitrogen-N

SON varied significantly between monthly samples (REML p<0.001, Figure 5.4.5), but there was no obvious seasonal pattern. SON was higher in 1999 and 2001 than in 2000 (REML p<0.01, Figure 5.4.5), although the lack of an obvious seasonal pattern makes the data hard to interpret. There was no direct effect of vegetation type on SON, but on some sampling dates, SON was higher under clover, especially early in the growing season. SON under grass patches was more variable than SON under clover (REML p<0.01, Figure 5.4.5). SON was not significantly different in different ages of ley, different rotations, or different management regimes.



Figure .5 Soluble organic nitrogen under grass and clover microsites in all ley plots at Tulloch, between February 1999 and August 2001. Points represent means of 14 samples. Bars represent standard errors.

4. Total soluble N

TSN was higher in 2000 than in other years, but the difference between years was not very large (REML p<0.001, Figure 5.4.6). TSN was up to 50% higher in 3 and 4-year-old leys than in 1 and 2-year-old leys (REML p<0.001, Figure 5.4.7). TSN was lowest in April (approximately 20 kg N ha^-1) and highest in the autumn (approximately 35-40 kg N ha^-1 p=0.001, Figure 5.4.6). This seasonal pattern varied between years, and TSN was unusually low in August 1999 (p=0.001, Figure 5.4.6 and Figure 5.4.7). TSN was not significantly affected directly by management, rotation or vegetation type. During the summer TSN was up to 30% higher in clover microsites than in grass microsites (p=0.001, Figure 5.4.6).



Figure .6 Total Soluble N under grass and clover microsites in all ley plots at Tulloch, between February 1999 and August 2001. Points represent means of 14 samples. Bars represent standard errors.

 



Figure .7 Total Soluble N in soil in 1-, 2-, 3- and 4-year-old leys at Tulloch between January 1999 and August 2001. Columns represent means of 8 samples (1-, 2- and 3-year-old leys) and 4 samples (4-year-old leys). Bars represent standard errors.

5. P, K and Mg

Grass and clover microsites had similar levels of extractable P and Mg in August 2001. P and Mg were also similar in cut and grazed leys and did not vary significantly between different ages of ley or different rotations. K levels were significantly higher in grass microsites than clover microsites in August 2001 (REML p <0.001), especially in 2 and 3-year-old leys (REML p <0.01, Figure 5.4.8). K levels were similar in cut and grazed leys and did not vary significantly between different ages of ley or different rotations.



Figure .8 Potassium levels in soil from under grass and clover microsites in 1-, 2-, 3- and 4-year-old leys in August 2001. Columns represent means of 4 samples (1-, 2- and 3-year-old leys) and 2 samples (4-year-old leys). Bars represent standard errors.

4. Discussion

1. Hypothesis a)

Nitrate-N was highest in 3 and 4-year-old leys, and lowest in 2-year-old leys. In 1-year-old leys, nitrate-N was also high on some sampling dates. Nitrate did not therefore show a steady increase with age of ley, although it was clearly higher in older leys (Figure 5.4.2).

Ammonium-N levels on almost all sampling dates were highest in 4-year-old leys and lowest in 1-year-old leys. Ammonium-N was clearly increasing with age of ley (Figure 5.4.3). Soluble organic N showed no clear pattern with age of ley. Total Soluble N was highest in 3-year-old leys (Figure 5.4.7).

Overall, soil N was accumulating over the lifetime of the ley, although the low soil nitrate-N in 2-year-old leys does not fit this pattern. Benefits to the soil from a fourth year of ley were not obvious. Four-year-old leys often had lower soil N than 3-year-old leys, possibly because the high grass yield in the 4-year-old leys removed large amounts of N from the system, so an extra year of ley may not greatly improve soil fertility. It is also possible that denitrification and leaching removed large amounts of N from 4-year-old leys. This confirms the findings of Johnston et al. (1994), who studied the effects of 1–6-year ryegrass-clover leys on soil fertility, and concluded that increasing the length of the ley phase beyond three years did not significantly increase the availability of nitrogen or subsequent crop yields. Granstedt (1992) also found that when clover content was allowed to fall below 30-50% of ley biomass, as it did in the 4-year-old leys in 2000, then offtake of N in grass is likely to exceed N input by N fixation.

Other data from the site indicates that P, K and Mg showed no consistent pattern with respect to age of ley, when data from different years was examined (Figure 5.5.2, Figure 5.5.3, Figure 5.5.4). pH tended to decline with increasing age of ley (Figure 5.5.5). It was difficult to distinguish the effects of ley age from the effects of management: all 1-year-old leys were grazed and all 2 and 4-year-old leys were cut. Cut leys had composted farmyard manure spread on them in the early spring, and this would have affected their nutrient levels. However, there was no evidence of elevated N, P, K or Mg levels in the cut leys. Nitrate-N was significantly lower in cut leys than in grazed leys in early summer, and no other significant differences between the soil in cut and grazed leys were observed.

2. Hypothesis b)

There was little difference between the two rotations in terms of their levels of nitrate-N, SON or TSN, but on most sampling dates, there was higher ammonium-N in leys of the 66% ley rotation than in the 50% ley rotation. This was true in all ages of ley, and was not affected by any seasonal trends. There was no significant effect of cutting or grazing on ammonium-N, so it is likely that this represents a long-term effect of the rotations on soil fertility. The 66 % ley rotation had only two years of arable crops in each cycle, and only one year without clover. On the other hand, removal of N in grass in the 4-year-old leys greatly exceeded N fixation in 2000. If the values of N fixation and grass N yield observed in 4-year-old leys were typical of previous years, then the 66% ley rotation might ultimately be less fertile than the 50% ley rotation.

3. Hypothesis c)

Nitrate-N was higher under clover patches than under grass patches over all sampling dates, and especially in summer, confirming hypothesis c. Total soluble N was also higher under clover patches in summer, and SON was also significantly higher under clover patches than under grass patches on some sampling dates. There was no difference between different ages of ley in this respect.

Differences were also observed between grass and clover microsites in the levels of K in summer. It is possible that K becomes depleted underneath clover patches, over time while nitrogen is elevated. Faerge and Magid (2003) suggest that the supply of potassium is a key limiting factor for growth and N fixation by organic white clover leys, although this was based on studies of farms in Denmark, so it may not be representative of the soil conditions in Scotland. Low clover content in grass-clover swards has been linked to potassium deficiency (Evans et al., 1986). All this suggests that white clover has a higher demand for potassium than grass. The alternative explanation for this data is that clover plants may be more likely to colonise microsites that are high in nitrate-N and low in potassium. This latter hypothesis seems unlikely, as the differences between clover and grass microsites seem to increase over time.

Differences in pH, calcium and phosphate have also been measured between clover and non-clover microsites in a number of upland sites in North Wales. Not all of these differences were consistent across all sites. The strongest correlation seemed to be between clover and calcium, but this may have been due to the positive relationship between calcium and pH (Snaydon, 1961). It is not possible to distinguish cause and effect in most of these cases: nutrient levels or pH may be high or low because of clover, or clover may be growing in a particular place because of the soil conditions. In some cases though it is likely that clover is affecting the soil: Matthew et al. (1995) observed that clover microsites had greater soil aeration, higher grass N concentration, and greater herbage mass than microsites without clover. In this case, the most likely explanation is that clover is elevating soil N in its immediate vicinity. The rate of mineralisation of N is greater under clover, grass mixtures than under grass monocultures, probably because of the low C:N ratio of clover residues (Elgersma and Hassink, 1997). This increased mineralisation of soil organic N, by clover may be a more important source of nitrogen for grass than directly transferred fixed N, especially in soils rich in organic matter (Mengel, 1996), like the soil at Tulloch. Mineralisation of soil organic matter by clover may explain the discrepancies between N transfer as calculated by the isotope methods, and the N difference methods (Section 3.6.8). Taken together this evidence suggests that clover grows preferentially in microsites with particular soil conditions, and that it also modifies the soil conditions locally, for example by increasing the availability of soil N.

4. Hypothesis d)

Nitrate under grass patches was higher than nitrate under clover patches on some sampling dates in winter, but the differences were not significant. If cyclical replacement occurred over a single growing season, clover would invade low N microsites early in the growing season, and elevate the soil N levels over the summer, before being replaced by grass late in the season, and retreating to a few low N microsites. There were considerable variations between samples gathered on the same date, especially in grazed leys, probably as a result of dung and urine patches, and this may have masked changes in soil N caused by the clover. This method only sampled soil from randomly chosen clover patches, it did not measure changes in soil N under individual clover patches, or closely examine changes in the distribution of clover patches in the field. The invasion experiment (Section 5) examines changes in the distribution of clover under experimental conditions, and addresses some of these questions.

5. Total and Soluble N in organic soils

1. Annual variation of soil nutrients

Nitrate-N and ammonium were both lower in 1999 than in 2000 and 2001, although SON and TSN did not follow this pattern. TSN was significantly higher in 2000 than in other years. SON was not characterised in this experiment, so it is not possible to state how much of this dissolved N was available to grass and clover. Little is known about the ability of clover to utilise organic N (Caradus, 1990). As ammonium was highly variable, and some of this variation may have been due to daily variations in temperature, rainfall etc., nitrate-N probably provides the clearest picture of the levels of soil N available to clover. Other data from the site indicates that over the period of study, soil P levels were highest in 1999 and lowest in 1997 and 2000 (Figure 5.5.2). K, Mg and pH did not change significantly between 1997 and 2000 (Figure 5.5.4 and Figure 5.5.5).



Figure .2 Phosphorus levels in soil from 1-, 2-, 3- and 4-year-old leys at Tulloch. Columns represent means of 4 samples (1-, 2- and 3-year-old leys) and 2 samples (4-year-old leys). Bars represent standard errors (SAC data).



Figure .3 Potassium levels in soil from 1-, 2-, 3- and 4-year-old leys at Tulloch. Columns represent means of 4 samples (1-, 2- and 3-year-old leys) and 2 samples (4-year-old leys). Bars represent standard errors (SAC data).

 



Figure .4 Magnesium levels in soil from 1-, 2-, 3- and 4-year-old leys at Tulloch. Columns represent means of 4 samples (1-, 2- and 3-year-old leys) and 2 samples (4-year-old leys). Bars represent standard errors (SAC data).

 

 



Figure .5 pH levels in soil from 1-, 2-, 3- and 4-year-old leys at Tulloch. Columns represent means of 4 samples (1-, 2- and 3-year-old leys) and 2 samples (4-year-old leys). Bars represent standard errors (SAC data).

2. Seasonal variation of soil nutrients

Nitrate-N, ammonium-N and TSN were all highest in autumn and winter. This is probably because of uptake of soil N by actively growing grass and clover during the growing season. Elevated soil mineral N in autumn under ryegrass-clover set aside swards has been observed by Chalmers et al. (2001), in the absence of any manure or fertiliser applications, strongly suggesting that some of the elevated mineral N results from rhizodeposition by clover. Eriksen et al. (1999) have also observed elevated nitrate under grass-clover leys in winter. As in this study, nitrate levels were higher in winter in the 2-year-old leys than in the 1-year-old leys. SON did not follow this pattern, although there were large differences between SON levels on different sampling dates. SON was high between August and December 1999 and low between August 2000 and April 2001. It is not clear what caused these variations, it may have been due to variation in the rate of production of SON by plant roots, uptake of SON by roots, or mineralisation of SON by microbes. There were no obvious relationships between SON and rainfall or temperature.

3. Effects of management on soil nutrients