Introduction

Global access to nutritious food at all times requires restructuring of the agricultural sector to limit the sector’s environmental footprint and meet Sustainable Development Goals (SDGs) (United Nations 2023). In line with global goals for sustainable food production, the EU Green Deal aims for at least 25% organic farmland (European Commission 2021; Montanarella and Panagos 2021; Purnhagen et al. 2021). Since organic farming systems are associated with relatively more sustainable nutrient management, increasing its farmland will affect two other Green Deal goals; reducing fertiliser use by 20%, and decreasing 50% in agricultural nutrient losses by 2030 (Heyl et al. 2023). Although there are opposing views on the extent to which organic farming contributes to sustainability and circularity (Leifeld 2012; Tscharntke et al. 2021), this farming system is doing well in several sustainability domains, such as animal welfare, low pesticides and farm-scale nutrient circularity (Nowak et al. 2013; Röös et al. 2018). Nonetheless, organic farming still requires nutrient inputs through fertilisation to sustain and improve productivity (Cooper et al. 2018).

Phosphorus is one of the essential nutrients for food production (Cordell et al. 2015). The rise in food demand and the inaccessibility of soil P led to the continuous application of fertilisers (Tóth et al. 2014). P-containing fertilisers are required for both conventional and organic systems to sustain and increase crop yields (Röös et al. 2018). Phosphorus mineral fertilisers are produced from phosphate rock, a non-renewable, non-substitutable natural resource that is geographically constrained in a few countries across the world (Cooper et al. 2011). Furthermore, due to inefficient use, P is one of the nutrients that have exceeded the planetary boundary, as such, dependence on primary resource P fertilisers threatens the sustainability of agriculture in the future (Steffen et al 2015; Sporchia and Caro 2023). Nutrient recovery and recycling are fundamental strategies for closing P cycles but minimising the amount of P in the cycle is arguably even more efficient and environmentally sustainable (Schröder et al. 2010; Papangelou and Mathijs 2021). Therefore, understanding the amount of P required to replace P outputs is essential.

Approximately 90% of the phosphate used for EU fertilisers is imported (EU Commission 2016). The EU being one of the biggest P importers in the world has promoted policies that restrict P fertiliser over-application and protect water quality over the last two decades (Nesme et al. 2016; Montanarella and Panagos 2021; Sporchia and Caro 2023; de Ridder et al. 2012; van Dijk et al. 2016). These policies are made to promote nutrient circularity and reduce P use inefficiency. Organic farming, from its inception, is designed to promote nutrient circularity through managing nutrient inputs, cropping systems and soil resources (Løes et al. 2017). But as nutrients are exported through market crops and other sold agricultural products, organic farming systems still have to restore nutrients removed through product harvesting. This is particularly important for P, since it cannot be recovered via biological fixation like nitrogen, and failure to replace it will lead to deficits and significantly affect harvests in the long run (Ohm et al. 2017). Furthermore, organic farming is associated with relatively higher P user efficiency through promoting farm-scale P circularity via the use of organic fertilisers including livestock manure, composts and green manure (Cooper et al. 2018).

The EU organic farming regulation emphasises the importance of limiting off-farm inputs into the farming system, leaving organic production with limited P sources. Allowed P sources in organic farming include composted organic residues, manure from conventional farming, grounded rock phosphate, and a limited use of synthetic mineral fertilisers (Cooper et al. 2018). Synthetic mineral fertiliser use is not in line with No. Article 2(1) of Regulation (EU) No 2018/848 for organic farming, but some derogations exist under Article 24(1) of Regulation (EU) No 2018/848 (EU Commission 2023). Limited synthetic mineral fertiliser use is allowable in organic farming if their use in non-organic farming has been approved per relevant exceptions union law (Regulation (EU) 2021/1165). Consequently, raw phosphate rock (Ca10(PO4)6(OH,F,Cl)2), Tetracalcium phosphate and aluminium calcium phosphate (CaAl(PO4)2) are the synthetic mineral fertiliser P sources allowed in organic farming (Moeller et al. 2018; FiBL Swedish input list 2021; European Union 2021). In this study, we will use “mineral fertilisers” to refer to these inputs.

Despite the strong restrictions on mineral fertilisers in organic farming, this practice contradicts the aims of the Farm to Fork (F2F) strategy by supporting a linear flow of P resources (Cooper et al. 2018; Reimer et al. 2020; Purnhagen et al. 2021). Therefore, it is currently considered if more recycled P products might be approved for use in the organic sector as well, although the use of recycled P products might raise other concerns, related to sanitary or pollution issues (de Boer et al. 2018). Organic farmland has been steadily increasing for approximately a decade in the EU; from 2010 to 2019 the land area for organic agriculture increased by 61% (Willer et al. 2019). Still, less than 10% of all agricultural land in the EU is under organic farming (Lampkin and Padel 2023). More than half of the EU organic farmland area is in Spain (2.3 million hectares (Mha)), France (2.2 Mha), Italy (1.9 Mha) and Germany (1.6 Mha) (Trávníček et al. 2021). The total area under organic farming in the EU27 + UK covered 14.6 million hectares in 2017, of which 45% is dedicated to arable farming while more than half is to permanent crops and grassland (Willer et al. 2021).

The targeted increase in organic farmland in the EU might trigger an increase of the fertiliser and P needs of the farming system. To anticipate such an increase and given the specific requirements regarding P inputs and the development and consideration of novel P fertilizer products, it is important to understand current P needs in organic farming. This paper aims to quantify and map P removal through crop harvest and mineral fertiliser P use in organic farming in the EU. Showing the contribution of different crop groups to the total P removed in the EU member states aims to support understanding of the P needs of organic farming across the region. Also, P budgets are an appropriate tool to project soil P due to the positive correlation between the two variables (Zicker et al. 2018). Therefore, understanding P that needs to be “replaced” can support the implementation of measures that prevent gradual soil depletion which can lead to the decrease in biological nitrogen fixation and limit food production (Zicker et al. 2018; Reimer et al. 2020). We provide a spatially explicit analysis of total P removal and use, which provides a foundation for optimising regionally available P resources that are allowed in organic farming in the EU or the extent to which mineral fertiliser contributes to P supply.

Methods

Background

We map P demand in two ways: as outputs via crop harvesting in organic farming, and as inputs through mineral fertilizer use. Both are mapped at the level of NUTS2 regions (Nomenclature of Territorial Units for Statistics; basic regions for the application of regional policies) across the EU. Replacing the nutrients that are removed through crop harvesting is one of the significant measures to maintain soil fertility (Schneider et al. 2019).

Dataset

We used a dataset from the Farm Accountancy Data Network (FADN) with reference year 2017. The dataset was requested in March and received in July 2022. FADN data is the only source of microeconomic data that is based on harmonised bookkeeping and serves as a base of knowledge for evaluation surveys or scientific research. It is a sample survey made annually, carried out with a uniform methodology at the EU level. Since 2003 the sample has been selected randomly, in full compliance with the requirements of statistical representativeness. Although FADN is mainly intended to be an accounting data source, it has been used for identifying environmental characteristics including farm type; organic, conventional or transitioning, and stocking densities (Kelly et al. 2018; Rega et al. 2022). It provides information that facilitates comparison of farm aspects between member states or NUTS regions within member states. The dataset contains data on 82,224 farms from 239 NUTS regions across the EU + UK, of which 6493 are organic or in conversion to organic. These sampled farms are representative of 90% of the EU’s agricultural production (Kelly et al. 2018).

For our analysis, we selected the organic farms and farms under conversion to organic, as farms under conversion abide by the rules and regulations of organic farming in the EU (European Commission 2023). We use production quantity values (in Mg) of market crops to calculate P removed through crop production, and the quantity of P2O5 in mineral fertilizers used at the farm level to quantify the amount of P applied. The FADN dataset has a variable (SYS02) that reports the number of farms represented by each sample point. We used the values in this variable for upscaling to the entire population. We compared P removed in organic farming with P removed in conventional farming for the same crop types from the same dataset, from 23 EU countries, and performed the same methods for conventional farming. Romania and Malta were excluded from the analysis because in the dataset there were zero organic farms in these countries.

The dataset reports production quantity values for individual fruits and vegetables, as well as for total fruit (CTOTFRUT_PRQ), total vegetables (CTOTHORT_PRQ), and several sub-categories. We only used production quantities for total fruits and vegetables to avoid double counting. However, lentils, field peas, beans and sweet lupin production values are not included in the CTOTFRUT_PRQ column therefore, they were added individually. The production quantity of mushrooms and tomatoes was subtracted from the value of vegetables (CTOTHORT_PRQ) so that they could be added individually. All the fruits that were included individually in the FADN data were also counted in the fruits (CTOTFRUT_PRQ) category; therefore we did not include any of them individually for our analysis.

Quantifying P removed through harvesting of market crops

There is a broad range of estimations of crop P concentration (Pc) values in harvested crops in grey and peer-review literature. We used Pc values of dry matter yields from Panagos et al. (2022a). These values are based on a recent, consistent literature review of empirical studies on P removal by crops across Europe (Table 1). We calculated P removal also following the soil surface P budget approach used in Panagos et al. (2022a).

Table 1 Crops considered and variables used to determine P removed through harvest in organic farming in the EU
Full size table

Production quantities of crops that were determined to contribute to P removal from arable land were selected (Table 1). This included market crops and grasses. When Pc were not available, we used Pc values of other crops in the similar crop group. Since the FADN provides production quantities at standard humidity, we first converted FADN production quantity figures into dry matter production quantities.

$$Dry\;matter\;production\;quantity = production\;quantity\;\left( {standard\;humidity} \right)*dry\;matter\;content$$
(1)

After that P removal for each crop separately at each sample farm was calculated:

$$P\;removal_{crop, \, farm} \left( {{\text{Mg}}} \right) = (Pc/ \, 100)*dry\;matter\;production\;quantity$$
(2)

This was summarized into a total P removal per farm, as well as P removal for the specific crop groups listed in Table 1:

$$P\;removal_{farm} = \sum \, P\;removal_{crop,\;farm}$$
(3)

For each farm, the total P removal was calculated and upscaled to the total farm population

$$P\;removal_{population} = P\;removal_{farm} *Farms\;represented\;by\;sampled\;farm$$

And lastly, we aggregated the results as a sum of P removal per NUTS2 region.

Mineral fertilisers:

FADN reports the amount of P2O5 used in mineral fertilisers in Mg. Since we were analyse the amount of P removed, for uniformity we converted P2O5 to P.

$${\text{P}} = {\text{P}}_{{2}} {\text{O}}_{{5}} /{2}.{29}$$

After the conversion, the values were upscaled to all organic farms by multiplying the P values with SYS02. Lastly, the quantities were aggregated to NUTS2 regions.

Analysis and visualisation

All data manipulation and analyses were done in R studio version 4.2.1, using the packages dplyr, reshape, bind, xlx, csv, magrittr, raster and ggpubr. The R code is available through [link added upon publication]. Visualization was done in QGIS long-term release version 3.28. 8-Firenze. To support interpretation, P removal was calculated for groups of crops (cereals, roots crops, oil crops, fodder, fruit and vegetable, permanent crops and permanent crops).

Totals per country, for both P removed and mineral fertiliser P, were calculated to support interpretation. Also, average P removal per hectare per country was calculated for organic and conventional farms, using the crop areas from the FADN dataset. For mineral fertiliser, we also calculated the fraction of P removed that is offset by mineral fertilisers. Interval classes for each crop category depending on the range of P removal values by the crops were created. The classes are arranged in an ascending order and they have different ranges and values per class.

Results

Phosphorus removal

Across the EU + UK, 61,296 Mg of P is removed through crop production in organic farming, with an average of 5.2 kg ha−1. The minimum P removed per NUTS2 region is zero Mg from six regions in the Netherlands, Poland, and the UK, where there are no organic farms that produce any of the crops included in our analysis (Fig. 1). All NUTS2 regions in the northern EU have P removal within 0–5 kg ha−1, except one region in Denmark. Western EU has the highest P ha−1 in the Netherlands (> 35 kg ha−1) which is the max P removal per hectare across the EU, followed by Germany with 20–25 kg ha−1 per NUTS2 region P removal (Fig. 1). In the southern region, P removal per hectare NUTS2 region is more diverse, with lower ranges found in Portugal and Spain, Italy has more NUTS2 regions in the 10–25 kg ha−1 P removal than any other country while Bulgaria and Greece range between 0- 20 kg ha−1.

Fig. 1
figure 1

P removed through harvest per hectare of organic farmland in each NUTS2 region

Full size image

We did not consider the various physiological, climatic and farm management factors that may influence P content in crops as we used P content values found in Panagos et al. 2022a. Therefore, P removal through crops is mainly influenced by the land area under organic farming and the production quantity mass of the crops. It is then expected that the countries with high organic farmland area are amongst the countries with high P removed through crop productivity (Italy, Spain, Germany, France). Italy has the highest P removed through crop production, with 27,300 Mg, followed by Spain with 5000 Mg (Table 2). According to the dataset used in this study, Italy accounts for 44% of total P removed through crop production, and within Italy, Sicily contributes 34% to P removed in the country. Sicily is also the NUTS2 region with the highest total P removed across the EU with 9237 Mg and 20–25 kg ha−1.

Table 2 Amount of P removed by organic crop farms per country
Full size table

On average, 258 Mg (5.6 kg ha−1) of P is removed per NUTS2 region. The variability across NUTS2 regions is high, as shown by the standard deviation of 759 Mg. These variations are due to differences in crop groups across NUTS2 regions. The contribution of different crops groups to each country’s total P removal varies (Fig. 2). In Italy, P removal is dominated by fruit and vegetable production, contributing 57% of the total amount in the country (Fig. 2). Heterogeneity in the  P content of different crops imply that the variability of P removed is also due to differences in crop compositions and crop yield. Cereals contribute 35% to P removal, fruits and vegetables 32%, fodder crops 20%, permanent crops 8% and root crops 1%. In more than half of the countries considered in this analysis, cereals production constitute at least 50% of P removal (Fig. 2). Ireland has more than 90% of P removed through cereal production with 11 kg ha−1. Only the Netherlands, Czech Republic and Slovenia have less than 10% P removal via cereals (Fig. 2).

Fig. 2
figure 2

Distribution of P removed through harvest in organic farming over crop categories per country in the EU (%)

Full size image

Fruit and vegetables contribute to P removal in all countries except Lithuania (Fig. 2). Under fruits and vegetables, legumes contribute ~ 10% of the total P removed and ~ 4% to the total P removal by crops in the EU. Cyprus is the only country that has P removed through fruits and vegetables but does not have P a fraction attributed to legumes (Fig. 2). All member states have less than 20% P removal through root crops (Fig. 2), with the highest fraction in Belgium at 112 Mg, making up 18% of P removed in the country. In Cyprus, more than 90% of P removed is from permanent crops (Fig. 2), but admittedly, Cyprus has the lowest P removal of EU countries and the P removed through permanent crops is only 0.04 Mg (Fig. 2). Spain is the only other country that has more than 20% P removal via permanent crops (Fig. 2) with 993 Mg from Castilla-La Mancha, making up 38% of P removal in the country. Due to diversity in crop production and the fraction of organic farmland dedicated to different crop groups, there are considerable differences in the maximum amounts of P removed by each crop group and the spatial patterns. P removal through cereal production is more dispersed across the EU, ranging from less than 100 Mg to over 1500 Mg per NUTS2 region (Fig. 2). While fodder crops and fruit and vegetables also have the maximum P removal that exceeding 1500 Mg per NUTS2 region, the values are skewed across NUTS2 regions (Fig. 3). Oil crops and root crops have the lowest contribution to total P removal, with a maximum of 288 and 92 Mg per NUTS2 region respectively. Notably, these are the only crop categories where the highest P removal is not observed  in a southern EU country (Fig. 3). The province of Liège in Belgium has the highest amount of P removal through root crops, with 92 Mg marking the only instance a  highest total P removal of a crop category is found in a western EU country. Cereals, fodder crops, fruit and vegetables and permanent crops have the highest-class values in southern EU. Latvia has the highest amount of P removed through fodder (Fig. 3), with 1095 Mg, accounting for  two-thirds of P removal in the country (Fig. 2), and is the only north-EU country with  P removal falling in the highest class of any crop category.

Fig. 3
figure 3

Total amount of P removed by harvest of different crop categories per NUTS2 region (Mg). Note the different units for e and f

Full size image

Differences between organic farming and conventional farming P removal

The amount of P removed in organic farming equals 4.3% of P removal from conventional farming. On average 5.2 kg P ha−1 is removed from organic farming, in each country, and 13.1 kg ha−1 is removed in conventional farming. In all countries, except Cyprus, the amount of P removed per hectare in conventional farming is higher than that of organic farming (SI, Fig. 1). Removal from conventional farming ranges from 2.5 kg ha−1 in Ireland and 32 kg ha−1 in Italy (SI, Fig. 1). Phosphorus removal per hectare by cereals in conventional farming is approximately 30% more than the average, at 18 kg ha−1. Cereal production contributes significantly to P removal for both organic and conventional farming as it is prioritised in the EU, in both organic and conventional farming (Magrini et al. 2016). Therefore, it provides a good example of the differences between conventional and organic farms and is elaborated here in more detail. Cereal crops have a wide spatial distribution with at least 100 Mg of P removed in more than half of the NUTS2 regions considered (Fig. 3). In organic farming, P removed through cereal production is 10 kg ha−1 per country, two times the farming system’s overall P removal average. The per-hectare P removal through crop production in organic farming is 41% of P removed in conventional farming when considering all 34 crops in the analysis. For cereals, P removed in organic farming is 60% of P removed from conventional farming. The highest gap is found in Portugal, where P removed through cereal crops in organic farming is 44% of conventional farming P removal, and the lowest gap is in Croatia where P removed by cereals from organic farming is 96% of conventional farming (Table 3).

Table 3 P removal (kg ha−1) through cereals from organic farming and conventional farming, and the percentage of P removed in organic farming compared to conventional farming
Full size table

Ireland has the highest fraction of total P removal from organic farming attributed to cereal crops (Fig. 2b). It also has the highest per-hectare P removal through cereals from conventional farming, while per-hectare removal from organic farming cereals is 10 kg ha−1 and is equal to the EU P removal average through cereals. (Table 3). The Netherlands on the other hand is among the countries with the lowest fractions and amount of P removal through cereals in organic farming, but has a very high per-hectare removal, with 17.8 kg P ha−1 in organic farming and 27 kg P ha−1 in conventional farming. Italy has the highest total P removed through cereals but P removal ha−1 is just approximately 50% of that in the Netherlands. In all countries, P removed through cereals in conventional farming is higher than in organic farming.

Mineral fertiliser use in organic farming

Overall, organic farming in the EU used 17,334 Mg of mineral fertiliser P (Table 4), it is is 28.1% of the P removed through crop production. On average, 73 Mg (2.2 kg P ha−1) of mineral fertiliser P is used in organic farming per NUTS2 region and 667 Mg per country. Out of the 26 countries considered in this study, 21 countries have mineral fertiliser P inputs in organic farming. Italy unsurprisingly has the highest amount of mineral fertiliser P used with 9588 Mg, followed by Spain with 1416 Mg. In the north EU the average mineral fertiliser P inputs is 4.0 kg P ha−1 per country. There are only two NUTS2 regions in northern Europe (Denmark and Lithuania) that report any mineral fertiliser P inputs and the highest amount is 84 Mg from Lithuania. Most western EU NUTS2 regions report mineral fertiliser P use below 200 Mg while in the Mediterranean and the south, there is a wider spread and higher P use (SI, Fig. 3). There is a relation (r = 0.5, p > 0.05) between the amount of P removed through crop production and mineral fertiliser P used per NUTS2 region. The spatial pattern for mineral fertiliser P use is similar to P removal with high P amounts in Italy followed by Spain (Table 4; Fig. 4a). This positive correlation suggests that increasing organic farmland area and production will not only lead to increased P demands but a likely increase in P fertiliser use in organic farming under unchanged policies.

Table 4 Total quantity and average application rate of mineral fertiliser P in organic farming
Full size table
Fig. 4
figure 4figure 4

a Average mineral P fertilizer rate per NUTS2 region in organic farming. b Percentage of harvest P removal that is offset by mineral fertiliser P

Full size image

More than two thirds of northern EU NUTS2 regions have 0% P offset by mineral fertilisers, while in the south, the majority of NUTS2 regions have at least 10% of the P removed offset by mineral fertilisers. Estonia, Finland, Latvia and Sweden do not have mineral fertiliser P inputs into organic farming (Table 4). The maximum and the low (greater than zero) amount of mineral fertiliser P are both in Mediterranean countries, Italy and France respectively. Within Italy, once again in Sicily with 3132 Mg, which accounts for 33% of the mineral fertiliser P in the country (Fig. 4).

Out of the 115 NUTS2 regions that have mineral fertiliser P inputs in organic farming, 23 of them have mineral fertilizer P inputs that are more than 100% of the P removed through crop production, while most regions, 37 in total, apply between 10 and 50% of the P removed through mineral fertiliser P (Fig. 3b). The percentage contribution of mineral fertiliser P to crop removed P has no distinct spatial pattern (Fig. 4b).

Discussion

Harvest P removal in organic farming is less than 10 kg ha.−1

The average P removed through the production of crops in organic farming is 5.2 kg P ha−1 per country. There are no other studies with empirical data on P removal through crops from organic farming across the EU national or sub-national scale. The estimates we found are lower than the average values in other studies, e.g. 14 kg P ha−1 (Panagos et al. 2022a) and 12.6 kg P ha−1 (Muntwyler et al. 2024), which can be attributed to the fact that these studies were producing estimates for P removal from EU arable farming, specific to neither organic or conventional farming. The spatial pattern of P removal through crop production (Fig. 1) is also different from Einarsson et al. (2020) and Panagos et al. (2022a) who both showed that P removal per hectare is highest in the western EU and low in the south EU. These differences may be partly  due to the different distribution of area and fraction of land under organic farming in different member states, and the  generally high crop harvests in western EU compared to other parts of the continent (Schils et al. 2018). In this  study, the countries with large areas of organic farmland (Spain, Italy, France and Germany; Willer et al. 2019) are among the countries with high total P removal through crop production (Table 2), suggesting  that the area under organic farming is a reliable indicator for projecting P removal and therefore, an important consideration for informing large-scale P management strategies and perhaps, broad-scale fertilising policies. The north region of the EU has lower P removal per hectare in this study (Fig. 1), in line with previous studies (Einarsson et al. 2020; Muntwyler et al. 2024).

The fraction of land dedicated to different crop groups varies between organic and conventional farming. Fodder is the largest crop group in organic farming in the EU, with more than 6.5 million ha of organic farming arable land utilized for fodder crops (Willer et al. 2021). The Czech Republic has approximately 550,000 ha of organic farmland (15% organic share), but is among the five countries with the highest total P removed, because P removal is predominantly influenced by fodder (Fig. 2). More than 30% of organic agriculture fodder in the EU is  from the Czech Republic, and it contributes approximately 90% of the P removed in the country. At the EU scale, however, organic farming fodder contributes 20% to the total phosphorus removal (Fig. 2b), less than fruits and vegetables. This can be attributed to the low mean dry matter P content (0.273%) of fodder crops compared to both fruit and vegetables (0.425%), and cereals (0.355%). On average, 1.8 kg P ha−1 is removed through fodder crops in each country whereas 10.6 kg P ha−1 is removed through cereal production. Cereals are the second most important crop in terms of area, accounting for ~ 2 million hectares of organic farmland in the EU (Willer et al. 2021). This large area, higher yields and relatively high spatial dispersion of the crop group can explain the dominance of cereals in total P removal¢. Pulses are the third most important crop in terms of organic farming area. While P removal is higher than that of cereals, the area is only 20% of that of cereals. Nevertheless, the area of pulses might explain the importance of vegetables in country-level P removal. Also, the variation in biomass production per NUTS2 region may explain P removal to some extent (r = 0.9, p > 0.05).

Determining the definite P requirements for an agricultural system at the EU scale is challenging because of the geographic and physiological-specific factors including soil pH, soil type, climate conditions and soil organic carbon, that can determine P demand and use by crops (Blackwell et al. 2019; Recena et al. 2022). But P removed from agricultural soil through product harvesting is one of the critical components of P flow in agriculture (Wang et al. 2016). In the EU, P removal through crop production is approximately 2.4 Mt per year (Panagos et al. 2022a). Using the same methods as Panagos and colleagues, we find that P removal through organic crop production is approximately 61,300 Mg per year. This is 2.5% of all the P removed through crop harvesting in the region while organic agriculture produces 4.7% of the total crop production (Lampkin and Padel, 2022; Panagos et al. 2022a). On average, P removal through conventional farming is 13.1 kg P ha−1 (SI, Table 3), this is closer to the EU average P removal found by Panagos et al. 2022a & Muntwyler et al. 2024. On the other hand P removal through organic farming is more than 50% less than P removed through conventional farming (5.5 kg P ha−1; SI Table 3). Differences in P removal quantities between the two farming systems are a proxy for quantity  of crops harvested given our assumption that since the P content of dry matter for similar crops remains consistent regardless of the farming system. We find that P removed in organic farming is 4.3% of P removed from conventional farming, while average P removal per hectare for organic farming is approximately 50% of conventional removal. These results align  with several studies (Seufert and Ramankutty 2017; Seufert 2019; Connor 2022) that have investigated the yield gap between these farming systems and further underscores the potential for enhancing land use efficiency in organic farming. We mainly highlight the contrast between cereals in the two farming systems because of the prominence of this crop group in EU agriculture (Kelly 2019). Therefore, limiting the likely amplification of differences in P removal as a result of varying crop compositions.

Fertiliser P use in organic farming is ~ 30% of P removed

Organic farming is intended to promote nutrient circularity (Løes et al. 2017), yet ensuring a sufficient supply of P for optimal crop productivity remains challenging and negative P balances are observed on organic farms in several regions across the EU (Ohm et al. 2017; Reimer et al. 2020). Although soil P stocks are considered additional nutrient resources in organic farming (Friedel and Ardakani 2021),  full exploitation of this  P source  is complex and not yet comprehensively understood (Teng et al. 2020; Walsh et al. 2023). Organic fertilisers including manures, composts and digestates are a significant P source to organic farming system (Cooper et al. 2018; Løes et al. 2017), their application can increase labile P pool availability. Countries with the highest total P removed per country (Table 2, SI Fig. 2) are also among the countries with low soil available P (Ballabio et al. 2019; Panagos et al. 2022a, b) and low organic P input (Muntwyler et al. 2024). While there is a positive correlation between P removal and mineral fertiliser P use, it is statistically insignificant (p > 0.05). This suggests that the higher mineral fertiliser P use in southern EU countries may be explained by low organic phosphorus input and low soil P availability. Approximately 1100 billion kg of synthetic mineral fertiliser P is used in agriculture annually in the EU (EUROSTAT 2023). Mineral fertiliser P used in organic is 28% of the P removed (Section “Mineral fertiliser use in organic farming”) and only accounts for 1.6% of total mineral fertiliser P used across the EU. Nonetheless, phosphate rock efficiency is limited to soils with pH > 6 (Cooper et al. 2018), and the use of this resource in organic farming is contentious because it supports a linear flow of nutrients which contradicts the principles of circularity and environmental care on which organic farming is based. Our estimates provide a lower boundary for the integrating more sustainable P sources into the farming system, particularly where overall P budgets are negative, and elimination of mineral fertiliser P use in case of P surpluses.

The use of mineral fertilisers ranges from zero inputs in most northern EU NUTS2 regions, to offsetting more than 100% P removal in sparsely distributed NUTS2 regions from other regions (Fig. 4b). Several factors can explain the amount of mineral P fertiliser used in organic farming. In Hungary, where mineral fertiliser P use is exceeds 100% of P removal (Tables 2 and 3), the scarce availability of manure from organic farming (which is an important P source) likely contributes to  a strong reliance on mineral fertilisers (PREPSOIL EU 2023). Sweden, Latvia, Estonia and Finland are the other extremes, with zero P use from mineral fertilisers, this could be due to national regulations. In Sweden, the national regulation for cadmium concentrations in P rock to be used in agriculture is 44 mg kg−1 P2O5 (Ulrich 2019), while for organic farming under the KRAV scheme under which the vast majority of Swedish organic farms is certified, Cd concentration has to be < 27 mg kg−1 for P (FiBL/KRAV 2021). In Finland, mineral fertilizers P use may be influenced by the 95% compliance to the Agri-Environmental Programme that sets specific limits for P in agriculture (Amery and Schoumans 2014). These four countries, and Denmark, have also received special attention with regard to limiting agriculture-driven nutrient pollution to the Baltic Sea (Kyllmar et al. 2023; Heyl et al. 2023). Although Denmark has mineral fertiliser P inputs, it is only 0.6 Mg (Table 4).

Limitations of methods and uncertainties in results

In this study, we approximated the P demand of European organic farming through P removal via crop production. A full P budget for cropland considers P input and output. Output variables include crop harvest, residues, soil erosion and emissions. (Papangelou and Mathijs 2021). Although other variables also contribute to P removal, crop production is responsible for approximately 95% of the P removed (Panagos et al. 2022a). Therefore, our results capture the major P flow and can contribute to P management, by providing insight into the main P outputs that require balancing. Furthermore, due to the relatively high P efficiency in organic farming, P from residues also contribute to P inputs. There is a high level of heterogeneity between farms regarding how they handle residues, which complicates quantifying their contribution, and it is beyond the scope of this paper. Moreover, a considerable portion of the harvested crops will be fed to animals and might be returned to the system as manure (Hijbeek et al. 2019). Consequently, there is a potential of overestimating the P removal, especially in regions with a high removal from fodder crops, including Latvia and Slovenia (Fig. 2b). As there is little insight into international manure trade, it is currently not possible to account for this in P demand estimates at this scale. Nutrient contents in crops can vary depending on the farming system (Herencia et al. 2011). However, we did not distinguish between conventional and organic farming, mainly due to data availability regarding P content differences between different crops from these farming systems. This can affect the results, especially comparisons between conventional and organic farming P removal comparisons.

The FADN dataset is the only dataset that provides farm-level insight into crop production and P fertilizer use, but it has its limitations. First, it only considers commercial-scale farms and does not provide data on smaller farms. The thresholds for farm sizes and economic significance are determined by the member states (European Court of Auditors 2016). However, the number of farms and heterogeneity largely match the number of farms in target population  reported by Willer et al. (2021). Therefore, we consider our data to be in line with other EU organic farming statistics. On average, farms that are managed fully or partially organic are 41 ha while the average farm size in the EU is 16 ha (European Union 2023a; b). This might skew the comparison with conventional farms and might have skewed the data for countries dominated by small farms, such as Cyprus. Furthermore, it is important to note that there are some inconsistencies and differences between the statistical databases (FiBL, IFOAM, EUROSTAT, FADN) that have EU organic farming data, which can affect this study. EUROSTAT reports crop production per NUTS regions only for cereals, dry pulses and permanent grassland, and does not have data on German, United Kingdom and Latvia NUTS2 regions (EUROSTAT 2024). The other datasets do not report crop production at scales smaller than country levels. In addition to this, crop grouping differs between the datasets.

Analysis of the FADN data, specifically for Italy, indicates that there is a replacement of non-responding farms with farms that have a different economic size and/or type of farming (Mari 2020). Nonetheless, the number of organic farms from Italy is consistent with other organic farming statistics and the spatial pattern of P removal is also consistent with the spatial distribution of organic farms in the region. However, this could have led to an over/under estimation of Italian production quantities which can affect crop P removal values.

Implications for management and policy

Topsoil P budgets have been used to quantify nutrient balances, forecast P changes and evaluate fertilisation management strategies for several regions, using different datasets in the EU (Oehl et al. 2002; Zicker et al. 2018; Einarsson et al. 2020). While previous studies have considered several input and output variables, this study quantifies and maps two components of the topsoil P budget for organic farming; P removal through crop harvesting and mineral fertiliser inputs, contributing to the understanding of P needs for organic farming at EU level. P removal by crops provides an upper boundary of P that should be replaced and the amount of mineral P fertilizer used provides a lower boundary. Even though our estimates vary by a factor 3.6, this range as well as the underlying spatial patterns provide essential information for further shaping and operationalising EU organic agriculture visions and policies. Spatial patterns of fertiliser use can facilitate the evaluation of variables that most contribute to the amount used in the farming system. This could be a starting point for finding more sustainable and likely recycled P sources to eliminate any dependence on mineral fertilisers.

Spatial mapping of these quantities per NUTS2 can facilitate a more precise allocation of P resources, thus contributing to Farm to Fork goals of reducing nutrient losses and mismanagement. Another suggested way to reduce nutrient losses is to reduce livestock production, mainly considering nitrogen (Björkbom 2023). Livestock reduction along with increasing organic farmland to 25% will transform EU agriculture in ways that influence both the supply and demand of organic farming P sources. Other European studies have displayed little change in yields following decades of omitting P fertilisation (Zicker et al. 2018), and it has therefore been suggested that many European soils contain sufficient legacy P to supply (organic) farming with low P inputs for decades. This underscores the importance of regular soil P monitoring and measuring to facilitate the efficient exploitation of this resource; avoiding over or under-P fertilisation (Bargaz et al. 2018). Especially because rebalancing P in the soils to get to levels that are conducive to sustainable productivity after a deficit is exhibited through low quantity and quality harvests is hard and complex, and threatens food security (Reimer et al. 2020). Nonetheless, more research is still required to fully exploit the potential for legacy P as it is only efficient under specific conditions (Teng et al. 2020; Gatiboni et al. 2021). Perhaps organic farming could expand in ways that are explicitly tailored to make use of legacy P.

Increasing the portion of agricultural land under organic farming will also require a restructuring of the types of foods that are produced and ultimately influence diets (Barbieri et al. 2019). Under increased organic farmland in the EU, there will be more land dedicated to leguminous crops, because of their significance to nitrogen fixation (Barbieri et al. 2019). Other EU field studies have already demonstrated that dependence on nitrogen-fixing crops is associated with negative P budgets in organic farms (Zicker et al. 2018; Reimer et al. 2023). In this study, we observe that legumes account for ~ 10% of P removed in organic farming (Fig. 2; Supplementary data Table 1) and have a relatively high average dry matter P content (Table 1). Therefore, increased production will further contribute to increased total P removal across the region and could lead to negative P budgets. While farmgate studies have indicated that organic farms are associated with negative P budgets, at a broad regional scale the farming systems’ P budget is not yet understood. A broad-scale P budget is important to determine whether organic agriculture inherently has low P available for use due to limited sources or poor distribution of available resources within the EU. This knowledge may be used for efficient implementation of regulatory frameworks such as the fertilising products regulation. We contribute to this knowledge by providing estimates for two components of the organic farming P budget.

Conclusion

Using FADN dataset, we found that organic crop farming in the EU + UK removes ~61 300 Mg P annually. This study provides a broad estimate of the P required to offset P removal and facilitate balanced P budget for organic farming across the region. Land area under organic farming is the main determinant for total P removed per country and thus essential to consider as an indicator estimating P fertilising requirements.  We also show that ~ 30% of P removed in organic crop farming in the EU + UK is offset by mineral fertiliser, thus showing the minimum P quantities to be replaced by more sustainable P alternatives. Organic farming contributes very little (1.6%) to the total mineral fertiliser used across the EU, which aligns with the farming system’s principles of environmental sustainability. Given the dependence on organic P sources including manures and composts in organic farming, future research could map potential P supply at the same resolution and demonstrate the extent to which mineral fertilisers contribute to the overall P inputs into organic farming. This would further facilitate matching the supply and demand and effectuate regional nutrient circularity at the NUTS2 scale or more detailed; a prerequisite for balancing future matches between demand and supply.