Agricultural soil acidity and phosphorus leaching risk at farm level in two focus areas

ABSTRACT

Poland has the largest agricultural area within the Baltic Sea drainage basin and reducing the risk of phosphorus (P) and nitrogen (N) leaching from Polish soils to water is therefore essential. Increased acidity is known to reduce soil fertility and may trigger P leaching from non-calcareous soils. As part of advisor training, 25 farms each in Pomerania and north-western Mazovia were visited and 1500 ha arable soil, including 180 ha grassland soil, were monitored in 2013–2014. The soil was mainly coarse textured, but 25% of the Pomeranian farms were dominated by silty or clayey soils. More or less regular drainage systems were found on 20% of the farms, while 50% had simpler, older (>30 years) systems with a few single pipes. The farmers often used only ammonium sulphate or other acidifying N mineral fertiliser. Median pH on the Pomeranian farms, analysed in potassium chloride solution [pH_(KCl)], was 5.2 and liming was advised for fields on most (72%) of these farms. Soil P content, measured by double-lactate extraction (P_DL), was positively and significantly correlated (Pearson coefficient 0.57; p < .001) with soil pH and was generally higher for pig farms. Mean P_DL (P given in elemental form) tended to be lower on dairy farms and arable farms and was significantly lower (mean 51 mg P_DL kg⁻¹ soil) on mixed farms (with just a few cows and poultry) than on pig farms (mean 122 mg P_DL kg⁻¹ soil). Farm-gate balances indicated deficits of P and potassium (K) on many of the small mixed farms in Pomerania and the soil can be expected to be nutrient depleted. In contrast, the pig farms demonstrated surplus farm-gate P balances (mean 27 kg ha⁻¹). The P leaching risk is discussed relative to soil threshold values and to results from Swedish long-term field experiments.

KEYWORDS:

• Farm walks
• farm-gate balance
• phosphorus
• potassium
• self-evaluation
• soil test

Introduction

Poland is one of the most important countries for agriculture in the European Union (EU), for example, it is the third largest producer of cereals in the E28, and is also largely water deprived (EUROSTAT 2015). However, Poland has the largest agricultural area within the Baltic Sea drainage basin (e.g. Rydén et al. 1993) and mitigating phosphorus (P) and nitrogen (N) leaching from Polish soils to water is therefore essential in order to counteract eutrophication of the Baltic Sea. Besides problems with low soil fertility, which reduce crop yield (Csathó & Radimszky 2012), Poland has issues of increased soil acidity, which has been demonstrated to cause enhanced P desorption at pH <6 from non-calcareous and medium- or fine-textured soils (Gustafsson et al. 2012). Increased risk of P leaching may consequently accompany low soil pH, especially if P uptake is low in a poor-yielding crop. Dystric Glossic Retisols, an acid, low fertility soil (WRB 2015), is frequently found in Polish lowlands. In sloping landscape, Stagnic Luvisols and Eutric Cambisols are common (Świtoniak 2014; Świtoniak et al. 2015; Waroszewski et al. 2015). For soil classification, pH can be measured in 1 M potassium chloride solution [pH_(KCl)], for fast results with a pH electrode (Brady & Weil 1996), a method also adopted for topsoil monitoring at farm level in Poland. For soil samples at laboratories involved in nationwide monitoring, pH can instead be measured in soil solution (1:1) with distilled water (USDA 2015).

The presence of artificial drainage systems is another important factor for P leaching. Such systems may alter water quality as a result of changes in hydrology and stabilisation of the groundwater level (Skaggs et al. 1994). Working tile drains change soil hydrology by improving water infiltration over a larger soil surface than in undrained soil. Improved drainage is generally followed by moderated P concentrations in water (Turtola & Paajanen 1995; Simard et al. 2001). Filtering of soil particles in infiltrating water together with less soil surface erosion may be other vital reasons for field drainage (e.g. Heathwaite & Dils 2001).

After long-term high P fertilisation, the soil P status can be expected to be enhanced (Saarela et al. 2004) but accumulation of P in agricultural soils may also follow from inefficient P uptake by the crop, for example, caused by deteriorated soil structure. Soil P tests, originally aimed at assessing plant-available P, are used worldwide for risk assessment of P losses from arable land, including the acid soils of northern Europe and Baltic states (Eriksson et al. 2013). The so-called double-lactate method (P_DL) according to Riehm (1942), with extraction at low pH (3.60) and a high (0.02 M) concentration of calcium lactate, is in use, for example in Poland and Latvia. Plant-available potassium (K_DL) is analysed after extraction with the same solution. A similar extraction method according to Egnér et al. (1960) with the calcium ions exchanged for ammonium ions at lower (0.01 M) concentration has similar low pH (3.75). This method (P_AL and K_AL) is routinely used in, for example Lithuania, Norway and Sweden and, with minor modification, also in, for example Hungary. The final P concentration in the extract is either analysed at high temperature by inductively coupled plasma mass spectrometry (ICP-MS) (Sweden) or spectrophotometry (Poland). The two nutrients are expressed as P₂O₅ and K₂O in southern Baltic regions but as single elements in most other countries. To further evaluate the P leaching risk, more or less readily dissolved forms of elements involved in sorbing P, iron (Fe) and aluminium (Al) can be analysed in the same extract. In the USA, another strongly acid extract (Mechlich 3) with pH 2.75, containing 0.27 M ammonium salts together with the complex-forming compound ethylenediaminetetraacetic acid (EDTA) is in use.

Soil P and K mapping is a planning tool for nutrient management by farmers. Indications of expected changes in soil P status are important from both an environmental and yield perspective. However, on fields with slopes and depressions, repeated soil sampling and soil P tests may be resource demanding. Phosphorus efficiency, estimated from a farm-gate balance, is another approach to indicate the direction of change in soil P_AL concentration, since approximately half the P surplus ends up in the soil in this form according to Swedish long-term fertility experiments (Eriksson et al. 2015). There are many international studies on N farm-gate balances, especially for organic and dairy farms, but few on P and K balances, especially in connection with soil mapping. Many farms with different types of production need to be involved to indicate any general trend, since history and soil management may vary widely between farms.

Knowledge of agricultural nutrient flows, soil quality and soil hydrology is essential for estimating the risk of high P leaching from agricultural land. Assessment of soil quality and observations of soil hydrology also form the basis for farmers’ own self-evaluation of the impact of their production on water quality. In the present study, two focus areas in northern Poland were studied with the aim of determining the risk of high P losses to water at farm scale. In order to facilitate comparisons with data from long-term field experiments in other neighbouring countries with similar soil and climate, a set of soil samples from a large pig farm was tested using the pH_(KCl), and P_DL methods at a Polish laboratory and using the P_AL method at a Swedish laboratory.

Materials and methods

Comparison of pH and soil phosphorus analysis methods

A total of 151 soil samples, 121 from topsoil (0–25 cm) and 30 from subsoil (25–60 and 60–90 cm depth), were taken on a large pig farm (RT-90 57900, 1550000) with undulating topography in West Pomerania Province, northern Poland, in summer 2014. The samples represented hills, slopes and depressions (322 ha) and for the topsoil a sample density of 0.38 ha⁻¹. Most topsoil samples were categorised as ‘light’ (82% of samples), ‘medium’ (10%) or ‘very light’ (8%), based on the share of particles <0.02 mm after suspension (Szulc 2002). Based on parallel particle analysis of sand, silt and clay fractions (mean 69:17:14) with the pipette method (ISO DIN 11277 1998), the soils were classified according to USDA (2015) as loamy sands, sands or sandy loams, which comprised 70, 25 and 30% of samples, respectively. Analysed in a corresponding way, subsoils were mostly (67%) categorised as loamy sands and sandy loams, with the remainder sands. The topsoil texture classification demonstrated no clear relationship to topography. All soil samples were pre-treated in accordance with ISO 11464 (1994) before analysis of soil pH_(KCl) (PN-ISO 1997), (ISO 10390:2005) and P_DL (PN-R-1996). These analyses were performed at the IUNG Laboratory, Soil Science and Plant Cultivation, National Research Institute, Pulawy, Poland. In parallel, all samples were analysed for P_AL (SS-ISO 028310 1993 and SS-ISO 028310T1 1995) at a commercial Swedish laboratory (Agrilab, Uppsala) frequently used for farmers’ own monitoring work. These samples were also analysed for Fe and Al in the same extract (Fe_AL and Al_AL) according to Ulén (2006). In order to sufficiently cover low soil P concentrations, another 41 loamy soil samples from Latvia and Lithuania (Eriksson et al. 2013) were used in the evaluation of soil P method.

Soil sampling and analysis on 50 farms

On 50 small to medium-sized (6–98 ha) farms (Figure 1) in two focus areas (Pomerania and north-western Mazovia, Poland), the topsoil of each field was sampled in autumn 2013 and spring 2014, giving in total 853 samples. Each was a composite sample representing 20 subsamples of mineral soil or 40 subsamples of grassland soil, together with a few organic soils. All soil samples, which were chosen to represent only one crop, were taken manually with a soil sampler in a cross, square or zigzag pattern (PN-R-04031:1997) and sent to one of three chemical-agricultural laboratories, the Regional Agrochemical Stations in Gdańsk and Kielce and the CDR Brwinów Branch, Radom. In total, 1500 ha arable soil, which included 180 ha grassland soil, were monitored, with a mean sample density of 0.57 ha⁻¹. All soil samples were pre-treated in accordance with ISO 11464 and analysed for pH_(KCl) (PN-ISO 10390:1997). The mineral soils were analysed for P_DL (PN-R-04023:1996), including spectrophotometric determination (ECS 1996) and K_DL (PN-R-04022 1996). The organic soils were extracted in 0.5 M HCl (PN-R-04024:1997) before analysis of P and K in accordance with Polish standards.

Figure 1. Sites of the 25 farms (filled triangles) studied in each of the two focus areas, Pomerania and north-western Mazovia in Poland.

Joint farm walk

As part of advisor training, each farm was visited three times and on one of these visits a joint walk was held, during which systematic observations were recorded (Ramnerö 2015). The farms represented different types of production (Table 1). According to the Polish method of soil texture determination, the dominant soil texture was ‘medium’ (43% of evaluated agricultural area) or ‘light’ (approx. 40%), with a small proportion of ‘heavy’ (13%) and a very small proportion of ‘very light’ (approx. 1%) soils, based on the share of particles <0.02 mm after suspension. Besides, a small proportion (slightly more than 2%) represented organic soils (Pietrzak 2014). These results were roughly interpreted in accordance with USDA nomenclature and are presented as mean value for each farm in Table 1. Field topography on each farm was assessed from internet maps (http://geoportal.gov.pl) showing elevation distance for each metre. The general topography was also appraised when walking on the farm. Hydro-technical parameters such as presence of artificial drainage systems and the farmers’ own reports of any recurring ponded water conditions were recorded, as were soil tillage practices. Number of animals expressed as livestock units (LIU) and farm-gate balances were estimated according to Pietrzak (2013), based on information given by the farmers to the advisors.

Table 1. Farm code, farm area (A, arable land), dominant soil texture class, existence of slope (S), artificial tile drain system, observations of water-ponded conditions (Pond. water) on the fields, usual season (spring or autumn) for ploughing before spring crops, livestock (LIU) density, mean topsoil concentrations of phosphorus extracted in double lactate (P_DL) and pH measured in potassium chloride solution (pH_KCl).

Code	A (ha)	Soil type^a	S^b	Drain syst.^c	Pond. water	Ploughing^d	Production^e	LIU (no ha^−1)^f	PDL (mg kg^−1)^g	pHKCl
RM01	18	Loam S	–	SP	–	Autumn	Pigs	2.7	171	6.9
RM02	88	Loam S	–	SP	Rare	Autumn	Pigs	1.2	45	6.7
RM03	30	Loam S	S	SP	–	Autumn	Dairy	2.4	68	6.2
RM04	38	Loam S	S	SP	–	Spring	Arable	0	71	5.7
RM05	34	Loam	S	SP	–	Autumn	Arable	0	154	6.8
RM06	55	Loam	S	SP	–	Spring	Dairy	0.8	110	6.2
RM07	34	Loam S	S	Sys	–	Autumn	Dairy	1.1	110	6.0
RM08	18	Loam	S	SP	Rare	Notill	Arable	0	88	6.2
RM09	20	Loam	–	Sys.	–	–	Horse	0.6	187	6.3
RM10	98	Loam	–	Sys.	–	Notill	Arable	0	124	6.1
RM11	40	Loam S	S	SP	–	Spring	Dairy	1.2	83	5.7
RM12	40	Loam S	–	SP	–	Autumn	Dairy	1.4	40	6.1
RM13	22	Loam	S	SP	Rare	Autumn	Arable	0	63	6.9
RM14	48	Loam	S	SP	Rare	Spring	Arable	0	108	6.8
RM15	86	Loam S	S	SP	Rare	Spring	Arable	0	48	5.7
RM16	45	Loam S	S	SP	–	–	Pigs	2.6	100	5.8
RM17	28	Loam	S	SP	–	Spring	Dairy	0.7	73	6.0
RM18	50	Loam		Sys.	–	Autumn	Arable	0	88	6.2
RM19	66	Loam	S	Sys.	–	Autumn	Arable	0	54	6.0
RM20	32	Loam		Sys.	–	Autumn	Arable	0	146	6.7
RM21	15	Loam		SP	Rare	Autumn	Arable	0	58	5.6
RM22	21	Loam S		SP	–	Autumn	Dairy	2.0	119	6.9
RM23	12	Loam S	S	SP	–	Autumn	Arable	0	66	4.3
RM24	69	Loam S		Sys.	–	Autumn	Dairy	0.8	83	5.3
RM25	18	Loam		SP	–	Autumn	Pigs	2.4	178	7.0
PM01	12	Loam S	S	–	–	–	Mixed	0.4	66	5.1
PM02	20	Loam S	S	–	–	Autumn	Dairy	1.6	67	4.9
PM03	13	Loam	–	–	–	Autumn	Dairy	1.0	49	5.4
PM04	16	Loam S	S	–	–	Spring	Mixed	1.2	51	5.1
PM05	21	Loam S	S	SP	–	–	Arable	0	39	4.7
PM06	20	Loam S	–	–	–	–	Mixed	0.5	91	6.0
PM07	29	Silt clay	–	–	–	Autumn	Arable	0	63	6.0
PM08	12	Loam	S	Sys.	–	Autumn	Dairy	2.5	71	4.5
PM09	9	Loam	S	–	–	–	Mixed	0.9	31	6.0
PM10	33	Loam S	S	–	–	Autumn	Mixed	0.2	45	4.9
PM11	19	Silt clay	–	–	–	Notill	Mixed	0.4	33	4.8
PM12	37	Loam S	–	–	–	Notill	Mixed	0.1	45	6.3
PM13	6	Silt clay	S	–	–	Autumn	Dairy	1.3	32	4.7
PM14	18	Loam	–	Sys.	–	–	Dairy	1.1	76	5.2
PM15	12	Loam	–	SP	–	–	Arable	0	18	5.7
PM16	16	Loam S	S	–	–	Autumn	Mixed	0.3	34	4.5
PM17	24	Loam S	S	–	–	–	Mixed	0.5	47	6.0
PM18	23	Loam	–	Sys.	–	Spring	Mixed	0.5	29	5.3
PM19	32	Loam	–	SP	–	Autumn	Mixed	0.6	51	5.6
PM20	18	Loam	S	Sys.	–	Autumn	Dairy	1.9	36	6.7
PM21	24	Silt clay	–	SP	Rare	Autumn	Mixed	0.9	58	5.6
PM22	30	Silt clay	S	SP	–	Autumn	Mixed	0.4	59	6.7
PM23	6	Loam	S	–	–	–	Pigs	1.6	116	7.2
PM24	41	Loam S	–	SP	–	Notill	Dairy	1.4	49	4.8
PM25	10	Loam	S	SP	Rare	Notill	Dairy	0.9	63	4.7

^a‘Light soils’ approximated to loamy soil dominated by sand (Loam S); ‘Medium soils’ to loamy soils (Loam) and ‘Heavy soils’ to silty and clayey soils (Silt clay).

^bS = Slope on at least some fields of the farm.

^cSP = single pipes; Sys. = more systematic pipe drainage; – = no drainage or no answer.

^dPloughing for spring crops; Notill = no tillage for any type of crop; – = no spring crop or not answered.

^eMain production; arable farms had only crop production; mixed had varied and low numbers of different animals.

^fRecalculated as for example one adult cow = 1; one piglet = 0.07; one hen and duck = 0.004; one dog = 0.05.

^gOrganic soils were extracted in HCl instead of double lactate.

Data analysis, statistics and estimates

All soil P and K concentrations given as oxides were recalculated to elemental form. The risk index ‘degree of P saturation in the AL extract (DPS_AL)’ was estimated on a molar basis according to Ulén (2006) for the large pig farm. All statistical analyses were performed in Minitab 16 (Minitab 2010). Mean and standard deviation were estimated for all parameters on each farm, for farms with the same type of production and for all farms in the Pomerania and north-western Mazovia focus areas. For soil P and K parameters, this comprised only samples representing mineral soils, with the limited area of organic soils excluded. Results from five farm-gate balances were also not included in the assessment, since these either represented a very special form of production (e.g. a horse farm) or were appraised as doubtful. The differences in soil pH, soil concentration (P and K) and use efficiency from farm-gate balances (P and K) between farms with different agricultural enterprises and the two focus areas were assessed with a two-sample t-test. Correlations (Pearson) were considered significant at p < .05 and the direction of any relationship evaluated with linear regression.

Results and discussion

Comparison of pH and soil phosphorus analysis methods

Inter-calibration (Figure 2(a)) of the two pH methods tested for the large Pomeranian farm revealed a 1.0 unit higher value with the method, but only in the range of 5.0–5.4 pH_(KCl). This difference is commonly generalised as a ‘rule of thumb’. In the range of 5.5–6.0 pH_(KCl), the difference was slightly less (0.9–0.8) units, while in a higher range ( 6.0–7.5), the difference was less again (0.8–0.5 pH units). Consequently, care should be taken when comparing results obtained with these two different pH methods at higher pH, at least for Polish soils.

Figure 2. (a) compared with pH_(KCl) for 151 soil samples from a pig farm. Coefficient of determination (r²) = 0.93. (b) P_AL compared with P_DL for soil from the same Polish (Po) farm, together with a set of (LiLa) Lithuanian and Latvian soils. The regression line corresponded to P_AL = 4.65 + 2.042P_DL, with r² = 0.89.

The inter-calibration between the P_DL method and P_AL method (Figure 2(b)) covering the P_DL range 15–430 (mg kg⁻¹, all values as P, not P₂O₅) with Polish, Latvian and Lithuanian soils demonstrated a regression line with the following equation:

Higher P_AL values were obtained than in a previous inter-calibration (Schick et al. 2013), which encompassed German and Polish soils (range of 75–359 mg P_DL kg⁻¹) but had no samples at all with less than ‘satisfactory’ P status (corresponding to P_AL < 137 mg kg soil⁻¹). Both datasets represented the same range of ‘slightly acid’ soils (measured as ) and the difference probably partly depended on different P solubility related to concentration and cation charge (Egnér et al. 1960). Another reason for the differing results is probably the final P determination with ICP-MS, which includes some organic P, unlike spectrophotometric determination (Maguire et al. 2005). The equation found in the present study had a high coefficient of determination (r² = 0.89). Soil P_DL was recalculated to P_AL according to this equation before being compared with results from P leaching studies in Swedish field experiments.

Phosphorus risk assessment on the large pig farm

Soil P concentration in topsoil from the large pig farm was high in terms of both mean and median values (Table 2). Soil P_AL values above 120 mg kg⁻¹ fall into the two highest classes (IVb and V) of six in Swedish soil nutrient assessments and are considered to have a risk of elevated P leaching (SEPA 2015). In all, 81% of the topsoil samples and 20% of the subsoil samples fell into these highest classes (Table 2). The DPS_AL threshold value of 30, which was set from degree of P saturation using Mechlich 3 extraction (Mechlich 1983) (DPS_M3) on US livestock farms (Butler & Coale 2005), gave similar shares of risk soil, namely 83% of topsoils and 26% of subsoils (Table 2). Those sites with high DPS_AL also in the subsoil should be regarded having a particularly high risk of P leaching (Ulén 2006).

Table 2. Total number (no.) of soil samples, phosphorus (P) in ammonium lactate extract (P_AL) as mean and median values, percentage of soil samples >120 mg P_AL kg soil⁻¹, corresponding to the Swedish soil P classes IVb or V, degree of phosphorus saturation in the AL extract (DPS_AL, %) as mean and median values and percentage of soil samples exceeding a DPS_AL threshold value of 30 in topsoil and subsoil.

	Topsoil	Subsoil
	(0–25 cm)	(25–60 and 60–90 cm)
Total no.	121	30
Mean PAL (mg kg soil^−1)	260	99
Median PAL (mg kg soil^−1)	210	78
PAL > 120 (% of soils)	81	20
Mean DPSAL	52	26
Median DPSAL	43	25
DPSAL > 30 (% of soils)	83	26

Topography, drainage system and tillage practices on farms in the two focus areas

The field topography was mainly flat, but with gentle slopes on most fields on 56% of the farms. There was no artificial drainage system on 30% of farms, while 50% had frequent single drainage pipes and 20% (mostly with clayey and silty soils) had full drainage systems at least in some fields. Only three farms had drainage systems younger than 20 years and the tile drains were on average more than 30 years old. However, nearly all were reported to function well. There were few visible signs of surface erosion on fields and only 16% of the farmers reported ponded water conditions in any field. Since few regular observations of turbid water were documented, and only in Pomerania, erosion problems seem to be limited to a few farms in this focus area. Soil erosion and notable slope processes have been reported for agricultural soils in the Brodnica and Chełmno lake districts of Pomerania (Świtoniak et al. 2015). In the present study, 16% of the farms performed spring tillage before sowing a spring crop, but late autumn tillage was the prevalent soil preparation method before spring sowing. Converting to spring tillage where possible is important in reducing N leaching, a conclusion also demonstrated with simplified N-leaching estimates based on Swedish experiments and available as an Excel sheet for advisors (Ulén et al. 2013). Moreover, even if visible soil erosion was a minor problem, frequent waterlogging, wet fields and flooding caused by poor management of watercourses were reported in pre-interviews with the farmers. Local and regional management of large ditches and streams therefore poses a future challenge in areas with flat topography.

Soil pH and phosphorus soil P status on farms in the two focus areas

Median pH on all Mazovian farms was 6.2 (corresponding to 6.9) and on all Pomeranian farms 5.2 (corresponding to 6.2), based on farm averages and analysed in potassium chloride solution [pH_(KCl)]. Liming was advisable for fields on most (72%) of the Pomeranian farms. Mixed farms typically had only a few cows and some poultry and were a frequent production form among the farms studied. For these generally small farms with low livestock density, many mineral soils had soil pH_(KCl) around 5.0 (Table 1). Many farmers applied acidifying N mineral fertilisers in the form of ammonium sulphate or urea as the main N fertiliser (in Mazovia 50% of farms and in Pomerania 70%), since other fertilisers were expensive. Soil P content was positively and significantly correlated (Pearson coefficient 0.57; p < .001) with soil pH. Acid soils were less common on pig and dairy farms than on arable farms (Table 1), demonstrating the neutralising effect of manure (Eghball 1999). Soil P_AL (mean 255 mg kg⁻¹ soil) on pig farms tended to be higher than on arable farms. Soil P_AL was significantly lower (mean 105 mg P_AL 100 kg⁻¹ soil) on mixed farms compared with dairy and pig farms. However, the direct relationship between number of LIU per hectare and soil P_AL was weak (Figure 3(a)) for these soils of unknown management history.

Figure 3. Soil P concentration on each farm estimated as extracted in ammonium lactate (P_AL) in Mazovia and in Pomerania, related to: (a) number of LIU, with only a very marginal correlation coefficient (p < .1) for the Pomeranian soils and (b) soil pH in potassium chloride solution. The two latter parameters were positively correlated (Pearson correlation coefficient 0.56; p < .001, coefficient of determination r² = 0.32). Points within the two circles represent farms with an elevated risk of high P leaching, that is, farms with very low pH or high soil P_AL concentration.

Swedish long-term fertility experiments have shown relationships between P_AL topsoil status and a risk of high P leaching, but this relationship has different shapes depending on soil type (Svanbäck et al. 2013). For some soils, the relationship increases exponentially above the 120–250 P_AL value. Topsoils with P_AL higher than 250 mg kg⁻¹ soil were suggested here as having a very high P leaching risk (Figure 3(b)). Including measurements of subsoil texture and DPS_AL might have given a more detailed risk assessment in a corresponding way to that of the large pig farm.

Swedish long-term fertility experiments at a high soil P level (equally high as on the Polish pig farms studied here) have demonstrated that further addition of manure increases P leaching even more than expected, possibly owing to a combined effect of manure P and soil P (Svanbäck et al. 2013). The mechanism for increased P leaching at low soil pH with fine-textured soils is unknown, but has been verified in Swedish long-term experiments (Gustafsson et al. 2012). Based on acidity and soil texture of the topsoil, there may be an enhanced risk of P leaching on 12% of Pomeranian farms, even if soil P status is simultaneously low. In view of the poor plant development on acid soils, such farms with coarser soil texture may also have a P leaching risk in principle (Figure 3(b)).

Farm-gate balances

Farm P surplus varied widely, and both positive and negative values were common, as found in this and other studies (Table 2). On all pig farms, there was a surplus of P and K and also of N, which varied between 200 and 340 kg ha⁻¹ (not given in Table 3). However, 31% of all farms were estimated to have a P deficit. Mean P surplus was highest for pig farms (mean area 42 ha) and lowest for mixed farms (mean area 23 ha) and this was also the case for mean soil P and K concentration. Farm-gate P balances on dairy farms were intermediate compared with values reported in Danish, Finnish, German and Portuguese studies (Nielsen & Kristensen 2005; Haas et al. 2007; Fanguerio et al. 2008; Huhtanen et al. 2011), and in a previous EU review (Swensson 2003). Mean farm-gate balances on mixed farms and soil P concentrations were low, as found in a previous Hungarian study (D'Haene et al. 2006, 2007). Potassium depletion may develop on 16% of the Polish farms studied here, especially on 35% of the mixed farms in Pomerania which had low import and export of nutrients.

Table 3. Mean (SD in brackets) phosphorus (P) and potassium (K) surplus from farm-gate balances, number (No.) of farms studied, animal livestock expressed as number of units (LIU) per hectare, mean (SD in brackets) soil P_AL and K_DL, country and reference. The studies encompassed one, three or several years. Soil P_DL (Polish) was recalculated to P_AL from the inter-calibration.

Farm surplus		No.	Animal	Soil parameters		Country	Reference
P	K			PAL	KDL
(kg ha^−1)			(LIU ha^−1)	(kg ha^−1)
Pig farms
15 (4)	–	19	1.5	–	–	Denmark^a	Nielsen and Kristensen (2005)
42 (8)	–	6	1.7	–	–	Denmark^b	Nielsen and Kristensen (2005)
27 (18)	77(17)	4	2.1	255 (152)	160 (55)	Poland^a	This study
Dairy farms
16 (4)	–	25	1.1	–	–	Denmark^c	Nielsen and Kristensen (2005)
7 (6)	–	13	1.5	–	–	Denmark^d	Nielsen and Kristensen (2005)
−3 (17)	1 (27)	26	1.4	–	–	Germany^c	Haas et al. (2007)
14−	–	–	1.3	100^h	–	Finland^e	Huhtanen et al. (2011)
46 (19)	100 (60)	8	4.7	220^i	212^i	Portugal^f	Fanguerio et al. (2008)
5 (25)	26 (70)	110	–	–	–	EU^g	Swensson (2003)
11 (14)	52 (48)	15	1.4	148 (61)	109 (29)	Poland	This study
Arable farms
3 (11)	–	7	0	100^j	–	Hungary	D'Haene et al. (2006, 2007)
3 (18)	49 (64)	14	0	162 (86)	49 (64)	Poland^h	This study
Mixed farms
0 (4)	–	7	–	100^j	–	Hungary	D'Haene et al. (2006 2007)
1 (9)	10 (14)	13	0.5	105 (42)	81 (50)	Poland	This study

^aIndoor pigs.

^bOutdoor pigs.

^cOrganic farms.

^dConventional farms.

^eModelled data based on several farms in SW Finland.

^fMedium intensive farms.

^gDairy farms with output from both livestock and crops.

^hRecalculated from a Finnish soil P routine method (P_AAAC) according to Schick et al. (2013).

ⁱHere recalculated from wet soils with a default density of 1.25 g/dm³. Potassium was measured as K_AL and was extracted in the same ammonium acetate acid as phosphorus (Egnér et al. 1960).

^jApproximate value from a graph.

Soil P concentration and P farm-gate surplus on the Polish farms studied were significantly positively correlated (p < .05) (Figure 4), as reported previously for 21 intensively managed dairy farms in SW Ireland with an annual deficit/surplus ranging between −3 and +47 kg P ha⁻¹ (Ruane et al. 2014). That study, which covered four years, also demonstrated significantly higher soil test P values (measured as Morgan's test) for farms with higher P farm-gate balances.

Figure 4. Farm mean soil P_DL concentration (mineral soils) and farm-gate P surplus/deficit (kg ha⁻¹ arable land) in 2013. Pearson correlation coefficient 0.46 (p = .02), coefficient of determination (r²) = 0.19. Soil P_DL tended to be higher in Mazovia than in Pomerania, but the difference was not significant.

Half the farms studied in Mazovia and Pomerania had a manure pad on which the manure was stored. The remaining farms stored their manure as compost in piles, either on the fields in which it is later spread or somewhere on the yard. Five of 12 arable farms in Mazovia imported manure from farms with animals and this resulted in a mean input of 12 kg P ha⁻¹, or 26% of the manure produced. In Pomerania, exchange of manure between farms was minor, even though manure pads and urine pits were more common there than in Mazovia. In general, more export of manure from pig farms and intensive dairy farms is needed in order to use the manure as a P source effectively and not build up soil P_AL to a higher level than at present on some farms and to avoid soil depletion on other farms.

Besides avoiding any future surplus P, complementary methods are needed to mitigate P leaching on livestock-intensive farms which have already reached high soil P levels. Such methods involve adding chemicals either directly to slurry (e.g. Brennan et al. 2011) or to the soil (O'Connor et al. 2005). Biochar has been suggested to be useful for increasing carbon storage and reducing climate change (Karhu et al. 2011), but contains much P and has been demonstrated to increase soluble P in soil water (Parvage et al. 2012). The use of magnetised biochar enriched in Fe (Chen et al. 2011) or the same amendment enriched in magnesium (Yao et al. 2013) might be a possibility to avoid this drawback.

In summary, acid soils are a problem for arable cropping and probably also to some extent for P leaching. Average soil P levels were moderate on the farms studied, but farm-gate balances indicated increasing P level for many farms with the highest soil P concentrations and depletion for small farms with low livestock density. More export of manure from pig farms to arable and small mixed farms is needed according to the farm-gate balances. These balances may be seen as an important tool for assessing environmental impact and can serve as an indicator of sustainable nutrient management. However, the analysis to be repeated in order to cover years with other production conditions (e.g. weather).

Acknowledgements

This study was funded by the private foundation BalticSea 2020, which is gratefully acknowledged. Many thanks also to Karin S. Tonderski, University of Linköping, Katarzyna Radtke, PODR, Ewa Strzeszewska, MODR, and Justyna Fila and Marek Krysztoforski, CDR, Radom, without whom this study would not have been possible.

Disclosure statement

No potential conflict of interest was reported by the authors.