Comparison of the NDICEA model's estimates for soil N with measured mineral N values at each of the trial sites

A RMSE of 20 kg N ha⁻¹ or less was proposed by Van der Burgt et al. (Reference Van der Burgt, Oomen, Habets and Rossing2006) to represent acceptable model performance for practical purposes. This could be achieved for most of the sites, although in some cases (e.g., EFRC and Hunts Mill) the modeled results are above this value (see calibrated model output in Table 1). The higher errors at EFRC could be a result of the small number of measurements (i.e., 7). The high errors for the stocked rotations at Hunts Mill could be explained by the fact that soil N measurements were taken soon after application of manures (the modeled values for this phase of the rotation were more than 100 kg N ha⁻¹ lower than the recorded values). It may also be possible that the model is underestimating the N supplied or that the deeper samples at Hunts Mill (0–60 cm) resulted in a mixing of topsoil and subsoil layers, and a subsequent overestimate of the mineral N content in the subsoil layer.

NPK balance for each of the rotations applied at the organic trial sites

Modeled nutrient balances derived from NDICEA are presented in Table 4 for each of the stockless rotations. The results include an estimate of the change in soil organic matter, with a negative number indicating mining of existing reserves and a positive number indicating an assimilation of N to soil organic matter (i.e., an increase in organic matter stocks).

Table 4. Nutrient balance of stockless organic rotations/trial sites expressed in kg ha yr⁻¹.

Table 4 illustrates that the amount of N supplied through biological fixation could potentially support the crop removal at the EFRC trial, although due to losses from the system, in particular leaching, much of this N is lost. The negative values for organic N indicated a mining of organic matter over the course of the experiment. This decline was observed through field measurements with organic matter levels dropping from 32 to 25 g kg⁻¹ of soil, probably as a result of starting the trial after a 5–6 yr ley. Organic matter levels would be expected to rise again on return to a longer-term ley period, as is a standard organic practice. The increase in organic matter levels from the implementation of a 1-year fertility building ley was reflected in the NDICEA model, which showed a rise during this period, although the subsequent decline was more than the offset gain. When a nutrient demanding crop was introduced into the rotation (e.g., potatoes in EFRC B) the deficit for all three nutrients increased to the extent that further use of external inputs (e.g., composts or manures) would be required, in particular for P and K. The high P and K offtake of beans similarly contributed to the large deficit of these two nutrients within the balance for rotation EFRC C.

A similar picture is presented for the rotation at ADAS Terrington. Despite the large contribution of N through fixation, there are considerable losses from leaching and denitrification, resulting in a negative N balance. It may be possible to address this problem through better use of over-winter cover crops (cover crops performed poorly over the course of the experiment with only one crop yielding over 1 t DM ha⁻¹). The deficits for P within this trial are also unsustainable in the long-term and would need to be addressed through imports or reducing the exploitative phase of the rotation. The K deficits for this site are also substantial; however, it may be possible that weathering of K stocks in the mineral pool could redress this (Khan et al., Reference Khan, Mulvaney and Ellsworth2014).

Lower rates of leaching were found for both Hunts Mill plots, although considerable deficits of P and K were also found despite the addition of green waste compost on area 1. The K deficits could be addressed through the buffering capacity of K-bearing minerals (Khan et al., Reference Khan, Mulvaney and Ellsworth2014); however, the P deficit represents a more fundamental issue for the maintenance of crop yields in the longer term (Cordell et al., Reference Cordell, Drangert and White2009). The results for Hunts Mill area 6 also illustrate that it is possible to maintain a fairly balanced system with regard to N through the effective use of late summer/autumn-sown green manures (i.e., without the use of a ley/break crop), although the overall deficit for N may result in a reduction in offtake or the need to use imported composts or manure.

Most of the stocked rotations were found to be more balanced with regard to N and P supply and loss (see Table 5). However, all of the SRUC sites faced a large K deficit, due to the high offtake from grass/clover silage, the potato crop (only at Woodside) and the use of straw for bedding, which were not offset by the manure application. Similar results were found within the nutrient balances for Tulloch and Woodside calculated by Watson et al. (Reference Watson, Younie, Stockdale, Cormack, Cook, Green, Holland, Leake and Welsh2000) although lower offtake was estimated within this study due to the lower assumptions on P and K content within NDICEA. Despite the K deficit, no trend in K levels was found over time at Tulloch or Woodside, although the soil samples were restricted to the first 30 and 45 cm due to the presence of indurated layers at deeper levels, largely impenetrable to soil augers or crop roots. It is possible that soil K levels at these sites were being supplemented by reserves within parent material, in addition to potential inputs from crop residues (these inputs are ignored in standard soil K measurements and in part explains why test values are unrelated to crop K balances; Khan et al., Reference Khan, Mulvaney and Ellsworth2014). These and other factors lead Khan et al. (Reference Khan, Mulvaney and Ellsworth2014) to suggest that measurements of available soil K are an unreliable indicator and that producers should use strip trials to determine site-specific fertilizer management. At both Woodside sites, the rate of N leaching was higher than at Tulloch, despite a lower annual rainfall. This was in part related to the lighter soil texture and the relatively low yield of the grass/clover leys at Woodside 37 (4–6 t DM ha⁻¹, Taylor et al., Reference Taylor, Younie, Matheson, Coutts, Mayer, Watson and Walker2006) which was related to the high soil moisture deficit. The low grass/clover yield at Woodside 37 also led to a negative N balance overall, due to a lower rate of biological N fixation. Much of the excess N at the other SRUC sites was locked up as organic matter (illustrated as a positive value under ‘Change in organic N’ in Table 5). This was observed within the trial through a small increase in soil organic matter levels observed at Tulloch, although the measured organic matter levels at Woodside remained relatively constant. Volatilization rates were low across all of the stocked rotations in Table 5 as a result of incorporating applied manure on the same day as application on the trial sites.

Table 5. Nutrient balance of stocked organic rotations/trial sites expressed in kg ha yr⁻¹.

Nutrient (NPK) balance for typical rotations applied using site conditions of the organic trials

Nutrient balances are presented in Table 6 for each of the typical rotations described in Table 3. The results presented below are mean values across all six sites and associated soil/weather conditions.

Table 6. Nutrient balance of typical organic rotations/trial sites expressed in kg ha yr⁻¹.

Note: se = standard error (se is only calculated for N as the P and K balances are unaffected by the site conditions).

The stocked complex rotation described above seems to represent a well-balanced system with regard to N and P supply and offtake. However, the model predicted a relatively large K deficit with offtake exceeding supply. As discussed earlier, this could be addressed through K delivery from the weathering of minerals depending on the underlying geology, climatic conditions and site management (Simonsson et al., Reference Simonsson, Andersson, Andrist-Rangel, Hillier, Mattsson and Öborn2007; Khan et al., Reference Khan, Mulvaney and Ellsworth2014), or through imported compost and/or mineral sources. The higher proportion of nutrient-demanding crops (e.g., potatoes) within the stocked simple rotation creates a larger K deficit compared with the stocked complex example. As with the stocked experimental sites in Table 5, volatilization rates were low for all of the stocked rotations in Table 6, as a result of selecting same-day incorporation of manure applied within NDICEA, and the low stocking density (i.e., 0.3 LSU ha⁻¹). The volatilization losses would be expected to increase if incorporation was delayed for any reason.

The stockless complex rotation has a deficit for all three nutrients. Two years of a red clover ley plus 1 yr of spring beans did not provide enough N to support 4 yr of crop offtake due to a high rate of leaching and denitrification. The presence of nutrient demanding crops contributes to the deficit (i.e., potatoes lead to a high N and K demand and beans to a high K offtake). The stockless simple rotation faces less of a deficit with respect to N, due to a higher input of biologically fixed N from the inclusion of peas which have a higher rate of N fixation than beans within NDICEA, and the use of the grass/vetch over-winter green manures following the spring crops. In addition, there is an absence of nutrient demanding crops (e.g., potatoes) however the rate of leaching is still high. The relatively low deficit of P within all of the typical rotations is a result of the application of rock phosphate. All of the modeled rotations would face a P deficit on a similar scale to the K balance without the use of this input.

Implications for improved organic management

In common with previous studies, the work presented here found considerable rates of N leaching within the rotations assessed (Berry et al., Reference Berry, Stockdale, Sylvester-Bradley, Philipps, Smith, Lord, Watson and Fortune2003; Torstensson et al., Reference Torstensson, Aronsson and Bergström2006; Kirchmann et al., Reference Kirchmann, Bergström, Kätterer, Mattsson and Gesslein2007). In some cases, this exceeded the amount lost by product removal (e.g., the stockless simple rotation described in Table 3). High rates of leaching under organic management are related to difficulties associated with matching crop N demand with N availability, particularly following incorporation of the ley when N availability exceeds demand (Torstensson et al., Reference Torstensson, Aronsson and Bergström2006; Aronsson et al., Reference Aronsson, Torstensson and Bergström2007; Bergström et al., Reference Bergström, Kirchmann, Aronsson, Torstensson, Mattsson, Kirchmann and Bergström2008). The use of organic manures can also make it difficult to predict N availability, compared with applications of mineral fertilizer (Cassman et al., Reference Cassman, Dobermann and Walters2002), making it more difficult to maximize N recovery and crop yields under organic management (Seufert et al., Reference Seufert, Ramankutty and Foley2012). As a result of these factors, lower use efficiency has been reported for organic cropping in comparisons with conventional systems (Torstensson et al., Reference Torstensson, Aronsson and Bergström2006; Bergström et al., Reference Bergström, Kirchmann, Aronsson, Torstensson, Mattsson, Kirchmann and Bergström2008).

The effective use of over winter green manures and undersowing of leys in cereal crops will help to reduce losses and thus enhance overall N efficiency (Kaffka and Koepf, Reference Kaffka and Koepf1989; Torstensson et al., Reference Torstensson, Aronsson and Bergström2006) and the lowest rates of leaching within this study were found for the rotations incorporating undersown crops and cover crops (e.g., ADAS Terrington, Warwick University Hunts Mill). Poor cover-crop establishment (e.g., at the ADAS and Elm Farm experiment) was experienced as a result of competition from weeds and slow emergence which reduced the benefit obtained (ADAS, 2006). Poor cover-crop establishment can also be related to competition from the cash crop, adverse weather conditions and low soil temperatures at the time of sowing (Snapp et al., Reference Snapp, Swinton, Labarta, Mutch, Black, Leep, Nyiraneza and O'Neil2005) In particular, the occasional occurrence of poorly performing cover crops presents an important challenge for the long-term sustainability of stockless systems, which rely on keeping the N supplied through biological fixation within the system. Although with careful rotation design, such systems are, in theory, sustainable from an N management perspective (Schmutz et al., Reference Schmutz, Rayns and Firth2007), in practice these systems appear to be highly vulnerable to poor establishment during the cover-cropping period.

The use of cover crops is not limited to organic farms, and higher N use efficiencies can be obtained by using this method alongside targeted mineral fertilizer application(s) to meet crop demand (and thus increase yield) whilst minimizing losses (Torstensson et al., Reference Torstensson, Aronsson and Bergström2006). Such tightly controlled systems could represent a suitable approach to developing highly N-efficient production systems, through a combination of organic practices and targeted fertilizer application (Cassman et al., Reference Cassman, Dobermann and Walters2002; Godfray, Reference Godfray2014). Similar targeted approaches could still be used on organically managed land, through the use of organic fertilizers with a high N availability (e.g., poultry manure and digestate from slurry-based anaerobic digestion) to supply readily available N at key points in the rotation (Berry et al., Reference Berry, Sylvester-Bradley, Philipps, Hatch, Cuttle, Rayns and Gosling2002; Möller and Stinner, Reference Möller and Stinner2009b). However, the application of such sources can increase the occurrence of nitrophilous weeds and their use within organic systems has been questioned as the high N availability leads to feeding the plant instead of the soil (Möller, Reference Möller2009a) and a reduction in the amount of organic matter applied in the case of digestate (Oelofse et al., Reference Oelofse, Jensen and Magid2013; Stinner et al., Reference Stinner, Möller and Leithold2008). The use of perennial crops can also help to reduce leaching in organic systems by keeping the soil covered and improving N synchrony (Di and Cameron, Reference Di and Cameron2002; Cox et al., Reference Cox, Bender, Picone, Tassel, Holland, Brummer, Zoeller, Paterson and Jackson2002), although lower yields, weed susceptibility and pest and disease management issues may limit uptake (Pimentel et al., Reference Pimentel, Cerasale, Stanley, Perlman, Newman, Brent, Mullan and Chang2012). A lack of technical information, suitable varieties and socioeconomic constraints (e.g., lower consumer demand compared with staple annual crops) also limit the potential for a wider adoption of perennial cropping (Pimentel et al., Reference Pimentel, Cerasale, Stanley, Perlman, Newman, Brent, Mullan and Chang2012; Valdivia et al., Reference Valdivia, Barbieri and Gold2012).

Organic farmers can also reduce N leaching considerably through improved management of manures and slurries. In particular, careful storage, application timing and choice of application method will help to maximize N recovery and minimize losses where slurries and manures are applied (Smith et al., Reference Smith, Beckwith, Chalmers and Jackson2002; Webb et al., Reference Webb, Pain, Bittman and Morgan2010). Manure analysis can also improve on farm nutrient use efficiency and help to reduce losses by improving understanding of nutrient supply from organic sources (Watson et al., Reference Watson, Öborn, Eriksen and Edwards2005). In some regions, there may be opportunities for farmers to work together to measure the nutrient use efficiency of their systems through a combination of manure and livestock dietary analysis combined with soil sampling (Verhoeven et al., Reference Verhoeven, Reijs and Van Der Ploeg2003; Le Gal et al., Reference Le Gal, Dugué, Faure and Novak2011). Such participatory approaches can be effective in allowing improvement options to be identified and for the fine-tuning of production systems. Again the use of such methods is not restricted to organic farms; however, the inability of such farms to access manufactured N fertilizer makes the implementation of such measures all the more important for the effective prediction of N supply.

With regard to P, the modeled systems able to achieve a sustainable balance were using external inputs of rock phosphate to offset losses. Although rock phosphate can help to offset losses, a reliance on this source may result in limited P bioavailability to meet crop demand, due to slow rates of solubilization (Edwards et al., Reference Edwards, Walker, Maskell, Watson, Rees, Stockdale, Knox and Lichtfouse2010). In addition, the use of such a fertilizer clearly does not fit well with the International Federation of Organic Agriculture Movements (IFOAM) organic principles (IFOAM, 2006), which emphasize the importance of reducing inputs to increase the long-term sustainability of farming systems. Despite this aim, the use of imported manure, straw and/or rock phosphate is common on organic farms, particularly for the supply of P and K (Kirchmann et al., Reference Kirchmann, Kätterer, Bergström, Kirchmann and Bergström2008; Oelofse et al., Reference Oelofse, Høgh-Jensen, Abreu, Almeida, El-Araby, Hui and de Neergaard2010; Nowak et al., Reference Nowak, Nesme, David and Pellerin2013). In many cases, manure and straw are sourced from conventional farms, which has led to the conclusion that organic farms are being ‘propped-up’ by conventional agriculture, and that as a result a large-scale conversion to organic management would be unsustainable (Kirchmann et al., Reference Kirchmann, Kätterer, Bergström, Kirchmann and Bergström2008; Oelofse et al., Reference Oelofse, Høgh-Jensen, Abreu, Almeida, El-Araby, Hui and de Neergaard2010; Nowak et al., Reference Nowak, Nesme, David and Pellerin2013). Organic monogastric systems (in particular poultry) also often require imported feed (e.g., soy) to supply protein and essential amino acids (Dekker et al., Reference Dekker, de Boer, Vermeij, Aarnink and Koerkamp2011) and so these systems are supplemented by internationally imported P and K.

The use of household waste and sewage sludge on organic farms could represent a possible solution to reduce the reliance on conventional manure and/or rock phosphate on organic cropping farms, in particular for the sustainable supply of P. Source separated urine also presents an opportunity to apply readily available N and P (Kirchmann et al., Reference Kirchmann, Kätterer, Bergström, Kirchmann and Bergström2008; Germer et al., Reference Germer, Addai and Sauerborn2011; Karak and Bhattacharyya, Reference Karak and Bhattacharyya2011). The use of such sources clearly fits with the organic ideal of closing the system as far as possible (Oelofse et al., Reference Oelofse, Jensen and Magid2013), although in this case the ‘system’ expands beyond the farm gate to the consumer (Oelofse et al., Reference Oelofse, Høgh-Jensen, Abreu, Almeida, El-Araby, Hui and de Neergaard2010). Although there have been many cases of household waste recycling on organic farms to supply nutrients to the soil (Darnhofer, Reference Darnhofer2005; Altieri et al., Reference Altieri, Companioni, Cañizares, Murphy, Rosset, Bourque and Nicholls1999; Luske and van der Kamp, Reference Luske and van der Kamp2009), the use of sewage sludge or urine is strictly prohibited on organic land in Europe, despite the fact that its use seems to be a rational and scientifically supported method of closing the nutrient cycle. Developments in the area of struvite (magnesium ammonium phosphate) recovery from waste water treatment plants could present a possible solution, allowing for application of a refined and slow release mineral fertilizer product, however this product is not currently on the list of permitted fertilizer within the European Commission organic regulation (European Council., 2007). This is an area that needs further scrutiny from a scientifically based perspective as it would appear that historical concerns about the toxic effects of applying urine and sewage sludge to agricultural land may no longer be justified (Smith, Reference Smith2009; Karak and Bhattacharyya, Reference Karak and Bhattacharyya2011), although public perception concerning the risks to human health remains an issue in some areas (Robinson et al., Reference Robinson, Robinson, Raup and Markum2012). Increasing the co-operative use of manure between (organic) livestock and arable farmers has also been suggested as a possible route for reducing the use of conventional manures on organic land and, within farming in general, the co-operative use of manure between specialized livestock and arable holdings could contribute to the prevention of stockpiling of nutrients and associated losses on intensive livestock holdings (Wilkins, Reference Wilkins2008; El-Hage Scialabba and Müller-Lindenlauf, Reference El-Hage Scialabba and Müller-Lindenlauf2010). In particular, this approach has been encouraged in Denmark by a decision to phase out the use of conventional manure and straw on organic land by 2021, partly in recognition of the conflict between principle and practice and partly to prevent the import of genetically modified organisms into organic systems via manure (Oelofse et al., Reference Oelofse, Jensen and Magid2013). In addition, the transition strategy in Denmark has highlighted the importance of crop rotation design (in particular to improve understanding on nutrient supply and losses), the development of crop cultivars for low-nutrient environments, and the development of biogas plants that can run on plant-based feedstock (in particular grass/clover harvested from leys) in recognition of the limited supply of organic manure (Jørgensen and Kristensen, Reference Jørgensen and Kristensen2010; Oelofse et al., Reference Oelofse, Jensen and Magid2013).

Potassium deficits were observed across all of the rotations however on many soils, this does not present an issue given vast reserves of mineral K within parent material which may be released for plant uptake by weathering (Khan et al., Reference Khan, Mulvaney and Ellsworth2014). Despite this potential, Holmqvist et al. (Reference Holmqvist, Øgaard, Öborn, Edwards, Mattsson and Sverdrup2003) found that weathering and bioavailability from the mineral fraction can vary greatly (between 3 and 80 kg K ha⁻¹ yr⁻¹ on a range of soil types in Norway, Sweden and Scotland), although the modeled predictions in this study did not take into account the dynamic and localized biological weathering by plant roots illustrated by x-ray diffraction studies (e.g., Hinsinger et al., Reference Hinsinger, Dufey, Jaillard, McMichael and Persson1991; Khan et al., Reference Khan, Mulvaney and Ellsworth2014) and the potential contribution of mycorrhizal fungi to K availability (Hoffland et al., Reference Hoffland, Giesler, Jongmans and Van Breemen2002). Nevertheless, improved knowledge of site-specific geochemical and mineralogical data, in addition to soil rhizosphere interactions could be a useful aid to develop site-specific fertilizer recommendations and nutrient balances (Holmqvist et al., Reference Holmqvist, Øgaard, Öborn, Edwards, Mattsson and Sverdrup2003; Andrist-Rangel et al., Reference Andrist-Rangel, Edwards, Hillier and Öborn2007). With respect to mineral reserves of K on the sites assessed in this study, only EFRC, IBERS and ADAS Terrington could be expected to supply a considerable amount of K from the clay fraction (Buckman and Brady, Reference Buckman and Brady1984), although sand- and silt-sized muscovite and biotite can also be a major source of plant-available K on lighter soils (Mengel et al., Reference Mengel and Rahmatullah1998) and the presence of these and other K-bearing minerals may have offset some or all of the K offtake at Hunts Mill and the Scottish sites (Andrist-Rangel et al., Reference Andrist-Rangel, Hillier, Öborn, Lilly, Towers, Edwards and Paterson2010). Despite the high deficits, there was no apparent trend in available K levels over time at most of the experimental sites considered, although Hunts Mill showed a slight decline over the course of the study and the K measurements at Tulloch (taken in the winter) may have been affected by the preceding silage crop (Watson et al., Reference Watson, Younie, Stockdale, Cormack, Cook, Green, Holland, Leake and Welsh2000). Other studies have demonstrated a decline in soil P and K levels following conversion to organic management (Loes and Øgaard, Reference Loes and Øgaard1997; Torstensson et al., Reference Torstensson, Aronsson and Bergström2006) and positive yield responses have been observed following K applications in long-term experiments in Australia and the UK (Bar-Yosef et al., Reference Bar-Yosef, Magen, Johnston and Kirkby2015) and within rice production systems, following several years of intensive cropping (Greenland, Reference Greenland1997). It is thus important to use nutrient budgets together with soil analysis to help understand the buffering capacity of soils and the management of P and K on individual fields. It should also be remembered that the bank-balance (i.e., supply minus offtake) concept of nutrient management can have major limitations, as N fertilization in excess of crop removal can lead to a depletion of soil carbon reserves by enhancing microbial decomposition (Khan et al., Reference Khan, Mulvaney, Ellsworth and Boast2007; Mulvaney et al., Reference Mulvaney, Khan and Ellsworth2009). This approach can also lead to an uneconomical fertilizer usage in the case of K that may also have an adverse effects on soil quality and productivity (Khan et al., Reference Khan, Mulvaney and Ellsworth2014) although a range of management factors (e.g., N supply and tillage system) can mask the effect of K fertilization on crop yield (Bar-Yosef et al., Reference Bar-Yosef, Magen, Johnston and Kirkby2015). It has also been suggested that crop yield and quality reductions following K fertilizer application are more likely to be related to K–Mg and K–Ca antagonism in plant uptake and/or K immobilization in the soil (Bar-Yosef et al., Reference Bar-Yosef, Magen, Johnston and Kirkby2015), rather than toxicity in the plant and root zone, or a depletion of the soil structure (Khan et al., Reference Khan, Mulvaney and Ellsworth2014).

In summary, it is clear from the analysis and modeling within this study that most typical organic cropping systems in the UK will require nutrient inputs to maintain an NPK balance. It should also be remembered that most organic farms import fewer nutrients than their conventional counterparts (Hansen et al., Reference Hansen, Kristensen, Grant, Høgh-Jensen, Simmelsgaard and Olesen2000; Torstensson et al., Reference Torstensson, Aronsson and Bergström2006; Trydeman Knudsen et al., Reference Trydeman Knudsen, Sillebak Kristensen, Berntsen, Molt Petersen and Steen Kristensen2006). Although this approach naturally leads to lower yields, and can lead to lower N efficiencies within cropping systems (Torstensson et al., Reference Torstensson, Aronsson and Bergström2006), it can also offer a useful way to balance production and environmental concerns (Gomiero et al., Reference Gomiero, Pimentel and Paoletti2011; Francis and Porter, Reference Francis and Porter2011). For example, organic farms often require less fossil energy on a per hectare or kilogram of product basis, in particular through the absence of imported mineral N fertilizer (Smith et al., Reference Smith, Williams and Pearce2015). The use of grass/clover leys and manures for fertility building on organic farms also contributes to greater soil organic carbon concentrations and stocks on organically managed land (Gattinger et al., Reference Gattinger, Muller, Haeni, Skinner, Fliessbach, Buchmann, Mäder, Stolze, Smith and Scialabba2012). In addition, organic methods (e.g., use of clover and other legumes to supply N) can be used effectively on conventional farms to increase efficiencies and reduce the environmental impacts of the agriculture sector as a whole (Pretty et al., Reference Pretty, Noble, Bossio, Dixon, Hine, Penning de Vries and Morison2005; Gaudin et al., Reference Gaudin, Westra, Loucks, Janovicek, Martin and Deen2013; Godfray, Reference Godfray2014).