The quantitative changes of nutrients in two contrasting soils amended with sewage sludge compost evaluated by various statistical tools

ABSTRACT

Introduction: The application of organic fertilisers to replenish soil organic matter and improve soil fertility and productivity has become common agricultural practice.

Aim of the study: This research deals with the effects of soil amendment with sewage sludge compost (SSC) on organic carbon, nitrogen total, nitrogen mineral and available P, K, S and Mg mineralisation in two contrasting soils. The various statistical tools used in this study have allowed us to present another conceptualisation of nutrient increments or losses as an effect of SSC applied. In order to distinguish groups of nutrients which are similar, a cluster analysis was used. A two-way analysis of variance was applied to compare the increments of the content of nutrients in the soils.

Material and Methods: A 3-year pot experiment was conducted, employing a randomised, factorial design with two soils (light and medium) and one amendment treatment as a compost at a rate equivalent to 6 Mg ha⁻¹. The following parameters of chemical soil properties were determined: contents of organic carbon (Corg), total nitrogen (Ntot), amounts of available P, K, Mg, S and mineral nitrogen (Nmin).

Results: The SSC showed a similar influence on the fertility of both soils. It was affirmed that application of SSC results in a statistically significant increase in the contents of soil organic carbon. The amounts of total and mineral N, as well as available P and S were subjected to different patterns of quantity changes expressed by both increase and loss. Moreover, a statistically significant loss of available K amounts was observed in both soils.

Conclusions: The findings of the study indicated that mature SSC becomes a long lasting fertiliser, slowly subjected to decomposition processes. Therefore, it may influence small increases in nutrient amounts in soils, in relation to the contents obtained for the control soil.

KEYWORDS:

• Sewage sludge compost
• soils
• nutrients
• cluster analysis

Introduction

Modern farming practices, such as intensive cropping, tillage and removal of crop residues, as well as decreasing manure and slurry application resulting from decreasing populations of farm animals, all contribute to the depletion of soil organic matter reserves and low nutrient availability in arable land. Among other things, these factors cause a deterioration of the soil environment and reduce soil fertility. So, to enhance productivity and restore degraded soils, fertiliser application is often necessary. Organic fertilisation is a way to substitute for inorganic fertilisers and improving general soil fertility. Thus, the use of composted sewage sludge amendments has become particularly important in the restoration of organic matter with significant amounts of essential nutrients (Mugnai et al. 2012; Jakubus 2013; Sciubba et al. 2014; Alvarenga et al. 2015). There are many papers dealing with the benefits of composted sewage sludge on different soil properties: physical, chemical, physico-chemical and biological (Bustamante et al. 2010; Duong et al. 2013; Jakubus 2013; Weber et al. 2014; Hernandez et al. 2016; Rigby et al. 2016). The use of composted sewage sludge is a very important strategy complying with the Landfill Directive (Council Directive 1999/31/EC) and with the ‘end–of–waste’ policy in Europe (Saveyn and Eder 2014). Moreover the use of such organic soil amendments also fulfils the postulate of the Thematic Strategy for Soil Protection (2006). In the case of compost application, the biological decomposition of organic matter should be taken under consideration. The mineralisation of compost may take place with different dynamics and in different ways and may affect soil properties. The effects of compost composition on the rate of compost decomposition and its influence on soil chemical properties is interesting, but quite well understood. Not surprisingly, most studies clearly demonstrate that soil amendment of compost enhanced with Corg, Ntot and available macronutrients (N mineral, P, K, S, Mg) results in a potential increased supply of plant nutrients. But in this study the evaluation of the fertilising effect of sewage sludge compost (SSC) on gain or decline in organic carbon and total nitrogen contents, as well as on the available amounts of N, P, Mg and K were examined. For the analysis of the data from this type of experiment, where we have a large number of traits, a cluster analysis is often used. This method allows the grouping of objects in homogeneous clusters. Cluster analysis is widely used in many fields of the natural sciences, medical sciences and genetic sciences (Zhao and Karypis 2005). The hierarchical method of cluster analysis is often used by researchers. This method can be visualised using so-called dendrograms (Aghili 2012). In environmental studies, non-hierarchical clustering methods, i.e. the k-means method, are very popular (Zou et al. 2015). Therefore, in this paper another approach to efficiency of sewage sludge compost application is proposed, based on the use of various statistical methods allowing the reliable evaluation of the real increment or loss in nutrients under controlled conditions of a 3 year pot experiment with respect to two contrasting soils.

Material and methods

Experimental layout, soils and amendment

A 3 year (2013–2015) pot experiment was conducted at the experimental station of the Agricultural University of Wroclaw. The two contrasting soils were used in the study. Samples of soils were collected from a depth of 0–30 cm from an agricultural field. The light soil was classified as Haplic Luvisols while the medium soil was classified as Haplic Cambisols according to IUSS Working Group WRB (2007). The soils were air – dried and passed through a 4 mm mesh screen. Commercially available compost was produced from a mixture of sewage sludge (50%), sawdust (20%), wood cuttings (10%) and wheat straw (20%) by the aerated-pile method. The basic chemical composition of the soils and sewage sludge compost (SSC) are presented in Table 1.

Table 1. Basic chemical composition of sewage sludge compost and soils used in the experiment (g·kg⁻¹).

Property	Sewage sludge compost	Light soil	Medium soil
Nog	22.6	0.53	1.17
Corg	545.87	6.6	15.31
P	4.96	0.116	0.269
K	3.47	0.218	0.412
Mg	4.34	0.028	0.150
S	1.89	0.029	0.056

The experiment was conducted in PVC pots (volume 10 kg) and employed a randomised, factorial design with two soils and one amendment treatment as compost at a rate equivalent to 6 Mg ha⁻¹ (0.2 kg of compost /pot) to comply with the limit of 170 kg N ha⁻¹ (EC 1991). According to this the scheme of experiment included: T0 – soil control and T1- soil with compost addition. 10 kg samples of dried soil were weighed in triplicate and mixed with a dose of compost. Each mixture was wetted to 60% field capacity. The plants were watered daily with tap water, or as needed, in order to maintain the moisture level. Three crops: white mustard, triticale and lupine were cultivated in the subsequent years of the experiment and were used as test plants. The results of the experiment regarding plant yield and quality will be published in another paper. In the first year of the experiment, soil samples were collected twice (1 month after compost application and after harvesting) and in the next 2 years – only once a year, after harvest. Soil samples were air dried.

Methods

The following parameters of chemical soil properties were determined: amount of organic carbon (Corg), total nitrogen (Ntot), amount of available P, K, Mg, S and mineral nitrogen (Nmin). The above properties were determined by methods commonly applied in soil science analyses (USDA 2014). Thus, organic carbon was measured using wet digestion with sulphuric acid and an aqueous potassium dichromate mixture, with the digests being back–titrated for residual potassium dichromate with ferrous sulphate. Total nitrogen was analysed by the Kjeldahl distillation method after wet digestion with sulphuric acid and selenium powder. Mineral N was extracted with 1 mol L⁻¹ NaCl at a 1:10 dry soil:solution ratio with 1 h shaking. The extracted solutions were analysed by the distillation method. Available P and K were extracted with buffered calcium lactate (Egner-Riehm method). Phosphorous was analysed colorimetrically as described by Murphy and Riley (1962). Available Mg was extracted with 0.0025 mol dm⁻³ CaCl₂ (Schachtschabel method) and available S was assessed turbidimetrically according to Bardsley and Lancaster (1960).

The chemical analyses of compost were conducted on dried samples. Total organic carbon (TOC) and total nitrogen (N_tot) were determined using Vario Max CNS. Total content of nutrients: P, K, Mg of compost were determined after incrimination at 550°C during 5 h and followed by digestion with 6 mol dm⁻³ HCl. The P content of compost was determined colorimertically using the molybdate-blue method (Murphy and Riley 1962). Concentrations of K and Mg in extracts were determined by flame atomic absorption spectrometry (FAAS) using Varian Spectra AA 220 FS. Total S of compost was assessed according to Butters and Chenery (1959).

Statistical analysis

For the analysis of experimental data, hierarchical and non-hierarchical methods of cluster analysis and two-way analysis of variance and Tukey method for homogeneous groups were used.

Cluster analysis is a tool of statistical data analysis which can be used for grouping objects in non-empty, distinct and possibly homogeneous groups called clusters. Objects belonging to a cluster should be ‘similar’ to each other, and objects belonging to different clusters should be as strongly ‘dissimilar’ to each other as possible. The main objective of cluster analysis is to detect in a data set so-called ‘natural’ clusters, and so clusters which can be reliably interpreted. The quantity of six nutrients in two contrasting soils was studied in this experiment. To confirm the well- known fact that the amount of elements in different soils are significantly different, the non-hierarchical, k-means cluster analysis method was used. The purpose of this method was to divide the data set into k clusters (groups).

A good division is one in which the sum of the distance of observations belonging to a given cluster is much less than the sum of observation distances between clusters. The k-means method consists in determining the coordinates of k points that are considered to be the centre of clusters. The observation will belong to this cluster whose centre is closest to it.

The aim was to use the k-means method for the random division of all nutrients with respect to amounts into two distinct groups - two clusters (corresponding to two different soils: light soil and medium soil).

In next step was the separate analysis of each type of soil, examining increments of the amounts of nutrients during the years of the study. Increments were calculated as the ratio of the content of an element in one year to the content of the element in the last year. And so, the increment marked for 2013 is the ratio of the content of the examined nutrient at the end of 2013 to the content of this nutrient observed at the beginning of 2013; an increment marked for 2014 is the ratio of the content of the examined nutrient at the end of 2014 to the content of this nutrient observed at the end 2013; finally, an increment marked for 2015 is the ratio of the content of examined nutrient at the end of 2015 to the content of this nutrient observed at the end of 2014. An increment equal to 1 means that the amount of the nutrient has not changed. An increment greater than 1 denotes an increase in the amount of the nutrient in relation to the previous year, an increment from 0 to 1 denotes a decrease in the amount of the nutrient, compared to the previous year. Increments ranging from 0 to 1 we will call a ‘loss’.

The method of k-means was again used to determine the fertilising effect of the applied SSC on soils (separately light soil and separately medium soil) i.e. whether the increment of the content of the elements on the fertilised soil were much higher (lower) in comparison to the contents of nutrients in the control soil. To confirm the results obtained by the k-means method, a two-way analysis of variance was used, taking as one factor – the dose of compost (occurring at two levels: zero dose – control soil, and soil amended with compost) and the year of study (2013, 2014, 2015) as the second factor. The analysis takes into account the interaction: dose x year. The trait measured was the increment of the nutrients amounts in the soil. Following this, the nutrients were grouped according to the values of the increment in the years of the study. For this purpose, a hierarchical cluster analysis was applied, and the results presented on dendrograms.

The cluster analysis made it possible to separate the groups of years that were similar to each other (or groups of nutrients similar to each other). Euclidean distance and Ward’s method of grouping was used, i.e. clusters of minimum variance were created. Then data from the soil fertilised with compost was analysed (separately for light soils and separately for medium soils). To discover whether these nutrients differed significantly with respect to the increment values, a twin-factor analysis of variance was used, taking into account the years and interaction. The first factor was assumed to be the nutrients occurring at seven levels: Corg, Ntot, Nmin, P, Mg, K, S, while the second factor was the year of research occurring at three levels (2013, 2014, 2015). The analysis took into consideration the interaction of the nutrient x year. In addition, the increment in all years of the experiment was taken into account. The measured trait was again the increment of the nutrient amounts. Following this, Tukey's method was applied to order to distinguish homogeneous groups on average over the years of the research. The significance of differences was marked: (*)- at α = 0.05, (**) at α = 0.01, (***) at α = 0.001.

Results

In this study the mineralisation effect of sewage sludge compost application on the fertility of 2 contrasting soils: light and medium was investigated. Applying the k-means method, including 2 groups, the amounts of analysed nutrients in both investigated soils amended with compost was presented in graphic form (Figure 1). The data in Figure 1 show the interaction between nutrients for individual soils and confirm common fact. The location of the dots (black and grey) in the rectangles informs about the amount of nutrient in the soils. The dots concentrated in the bottom left corner show small amounts, and the dots concentrated in the top right corner show larger amounts. The amounts of particular elements obtained for light soil (black) were lower in comparison to the amounts of the same elements but determined for medium soil (grey).

Figure 1. Contents of nutrients in light (black dots) and in medium soil (grey dots) presented in pairs (contents of Mg and P, contents of Mg and K, contents of Mg and S, … , contents of Ntot and Nmin). The dots concentrated in the bottom left corner (black) on each chart show small amounts of element (corresponding to light soil), and the dots concentrated in the top right corner (grey) show larger amounts of element (corresponding to medium soil).

Light soil

The analysis of the amounts of the individual elements for the two agronomic soil categories was conducted using the k-means method. In order to reliably assess the effect of compost fertiliser, further statistical analysis was based on the increments of content of the individual nutrients. A two-way ANOVA (factors: dose of compost and year of study) confirmed the results obtained by k-means method, namely, that the dose of compost did not affect the increase (value of increment) or decrease (value of loss) of nutrient amounts in the soil. In the analysis, the interaction of these factors was taken into account. The p-values of these comparisons for seven separate analyses (for each element separately) are displayed collectively in Table 2.

Table 2. The p-values of the results of seven independent analyses of variance (seven nutrients) for two type of soils (light soil and medium soil).

	C	Ntot	Nmin	P	K	Mg	S
Light soil
Compost dose	0.186	0.103	0.560	0.224	0.990	0.540	0.409
Year of study	0.295	0.106	<0.001***	0.168	0.004**	0.046*	0.0209*
Dose × year	0.111	0.005**	0.015*	0.924	0.063	0.310	0.069
Medium soil
Compost dose	0.009**	0.930	0.831	0.665	0.808	0.454	0.204
Year of study	0.001***	0.006**	0.014*	<0.001***	0.018*	0.097	0.217
Dose × year	0.497	<0.001***	0.025*	<0.001***	0.325	0.007**	0.405

Two-way analysis of variance for comparison of the increments of nutrient amounts in the soil fertilised with SSC (dose: 0.2 kg of compost /pot) to the control soil (dose: 0 kg of compost /pot) in the years of the studies (2013, 2014, 2015) and their interactions (dose × year) for seven nutrients.

Significance level: *p < 0.05, **p < 0.01, ***p < 0.001.

The data included in Table 2 show that none of the content of nutrients (with respect to the increments) found in the soil fertilised with SSC did not differ significantly from those obtained for the control soil. The impact of the second factor – the year of the study – proved to be significant for Mg amounts, highly significant for K amounts and extremely significant for Nmin. Besides, the increment or loss of magnesium, both in the fertilised and control soils, was significantly different. Similarly changes in the content (expressed by increments or losses) of K and Nmin in various years were significantly different. The interaction of the experimental factors was important for the amounts of Nmin and very significant for Ntot contents. The changes in the content of elements (after 2013, after 2014 and after 2015) obtained for the control soil (T0) and the soil enriched with compost (T1) were significant.

Using cluster analysis, all the years were divided into two clusters of similar years with respect to the increments of nutrients. From the dendrogram (Figure 2) it can be seen that the years are not very similar with respect to increments of nutrients, although we can observe that 2015 and 2013 form one cluster, and 2014 is separate. Thus the average values of the increments or losses of all studied nutrients in 2013 and 2015 were more similar than the average values of the increments or losses of all studied nutrients in 2014.

Figure 2. Dendrogram. Grouping of years (2013, 2014, 2015) with respect of the average values of the increments or losses of nutrient amounts found in light soil amended with sewage sludge compost (T1 – dose 0.2 kg of compost /pot) and control soil (T0 – dose 0 kg of compost /pot). Dark colour – large increase in the content of elements in the soil, light colour – decrease in the content of elements. The years 2013 and 2015 are slightly similar in respect of average increase or decrease in content (nutrients) in the soil.

Again, the grouping the contents of nutrients with respect to their increments or losses in the light soil fertilised with SSC was made using cluster analysis (Figure 3). They were divided into clusters characterised by similar changes in the element contents during the experiment. Figure 3 presents the trends in the increases in the contents of nutrients in the light soil amended with SSC in particular years of the experiment. On the basis of the data, increments (from 3% to 24%) of amounts were only found in the case of Corg and . The amounts of other analysed nutrients in particular years of the study suffered losses (increments ranging from 0 to 1). This trend was particularly visible for K, for which amounts in light soil after the 1st, 2nd and 3rd years of experiment decreased by 31%, 7% and 37% respectively. The Ntot amount after two years increased (by 7% and 16% in 2013 and 2014) and then decreased by 8% in 2015. The reverse tendency in quantitative changes was noticed in the available P. At the beginning of study, the P amount decreased (by 4% in 2013 and by 8% in 2014) followed by an increase (by 5% in 2015). A similar pattern of quantitative changes was found for the available Mg and Nmin, and this means that after the 1st and 3rd years there were losses (5%–17% for Nmin and 2%–19% for Mg). For the soil samples represented the last year of the experiment, increases (by 7% for Nmin and by 20% for Mg) of nutrient amounts were revealed (Figure 3).

Figure 3. Dendrogram. Grouping of elements with respect to average increments or losses of the amounts of nutrients across all three years (2013, 2014, 2015) in light soil amended with sewage sludge compost (dose 0.2 kg of compost /pot). Dark colour – large increase in the content of elements in the soil, light colour – decrease in the content of elements. The nutrients C and Ntot are similar in respect of average increase or decrease in content, Mg and Nmin are similar, and S and P are slightly similar in respect of average increase or decrease in content.

In order to assess the significance of the impact of compost SSC on increments of the contents of nutrients, the two-way analysis of variance taking into account two factors, namely SSC dose and the year of experiment, was conducted. Additionally their interaction for fertilised soil was taken into consideration (Table 3).

Table 3. The p-values of two-way analysis of variance for comparisons of increments of analysed nutrient amounts (compared nutrients: C, Ntot, Nmin, P, K, Mg and S) for two type of soils (light soil and medium soil) in the years of the studies (2013, 2014, 2015) and their interactions (Nutrient × year).

	Light soil	Medium soil
Nutrient	<0.0001***	<0.0001***
Year of study	<0.0001***	0.357
Nutrient × year	0.003**	<0.0001***

Significance level: **p < 0.01, ***p < 0.001.

The data in Table 3 indicate that changes in the contents of various nutrients differ very significantly between nutrients (p-value < 0.0001) as well as over the years of the study (p-value < 0.0001). The interaction of the experimental factors is also significant (p-value < 0.01). Since the differences in the increments or losses of nutrient amounts were statistically significant, homogeneous groups in the soil fertilised with compost were determined. According to the data presented in Table 4, three separate groups can be distinguish within the analysed group of nutrients defined by increment or loss of Nmin (group c), S (group a) and K (group d). The greatest observed loss was for the available K amounts, and the greatest increment was observed for the available S. The increments or losses for other nutrients were at comparable levels.

Table 4. The homogeneous groups of nutrients with respect of increments their amounts in Tukey test for two type of soils (light soil and medium soil) on average over the years of studies. (groups a, ab, abc are homogeneous; and groups a, b, c, d are separable).

Light soil
Nutrient	S	C	Ntot	Mg	P	Nmin	K
Group	a	ab	abc	abc	bc	c	d
Medium soil
Nutrient	C	P	Ntot	Mg	Nmin	S	K
Group	a	ab	ab	ab	b	b	c

Medium soil

An analysis of the increase in the amount of individual nutrients was conducted for the medium soil, too. All the data for the soil control and soil fertilised with SSC were studied. In order to distinguish two homogeneous groups of nutrients (one group in the control, and the second group in the soil amended with SSC) the k-means method was applied. For the medium soils, the effect of the compost on the quantitative changes of the nutrients under investigation was not significant. The results of the k-means method were confirmed by twin-factor (the dose of compost and the year of research) ANOVA analysis with regard to the interaction of both factors. The p-values of these comparisons for seven separate analyses (for each element separately) are shown collectively in Table 2. Based on the results shown in Table 2, it can be concluded that increases in the content of all the nutrients (except for C: p-value = 0.009) in soil fertilised with SSC were not significantly different from the increases obtained for the control soil. The impact of passing time in the conducted experiments on changes in the contents of analysed elements was significant in relation to the Corg and available P. Meanwhile, the interactive action of the experimental factors was significant for Nmin content, highly significant for the available amounts of Mg and extremely significant for P and Ntot contents (Table 2).

The cluster analysis, the results of which were presented in a dendrogram (Figure 4) revealed that the average values of the increments or losses of nutrients in 2014 and 2015 were more similar than the average increments or losses of nutrients in 2013.

Figure 4. Dendrogram. Grouping of years (2013, 2014, 2015) with respect of the average values of the increments or losses of nutrient amounts found in medium soil amended with sewage sludge compost (T1 – dose 0.2 kg of compost /pot) and control soil (T0 – dose 0 kg of compost /pot). Dark colour – large increase in the content of elements in the soil, light colour – decrease in the content of elements. The years 2014 and 2015 are slightly similar in respect of average increase or decrease in content of elements (nutrients) in the soil.

Again using the cluster analysis, the grouping of nutrients was carried out with respect to changes in their contents in medium soil fertilised with compost (Figure 5). The elements were divided into clusters characterised by similar increments (values above 1) or losses (values in the range 0–1) in the years of the study.

Figure 5. Dendrogram. Grouping of elements with respect to average increases or losses of the amounts of nutrients across all three years (2013, 2014, 2015) in medium soil amended with sewage sludge compost (dose 0.2 kg of compost /pot). Dark colour – large increase in the content of elements in the soil, light colour – decrease in the content of elements. The nutrients C and Mg are similar in respect of average increase or decrease in content, Nmin and S are similar in respect of average increase or decrease in content.

Just as in the case of the light soil, in the conditions of medium soil amended with SSC, an increase in Corg content was found (10% after 2014 and 12% after 2015). Again, the amounts of other analysed nutrients in the particular year of study underwent losses (increment ranging from 0 to 1). This trend was especially visible for K, for which amounts in medium soil after the 1st, 2nd and 3rd years of experiment decreased by 40%, 23% and 28% respectively. A similar pattern of quantitative changes, although to a lesser degree, was noted for the available S and Mg. Their amounts in the soil reduced after each year of experiment by 2%–21% (S) and by 1%–4% (Mg). The data presented in Figure 5 indicated a significant increment of total nitrogen in the soil after the 1st year of study (by 34%), however, the following years revealed an evident loss of nutrient content by 18%–19%. The reverse tendency was observed in the amounts of mineral N in the soil in particular years of the experiment. Soil amended with SSC showed lower nutrient levels of 11% and 18% after the 1st and 2nd years, and then a 4% increase was noticed. A considerable increment (by 38%) of available P amounts after the 2nd year of study was established, however the soil samples representing the last year showed a decreasing tendency and nutrient amounts were lower by 21% (Figure 5). Similarly as in the case of the light soil, in conditions of medium soil, the amounts of elements were compared using a two-way analysis of variance performed only for the data specified in the conditions of the soil fertilised with SSC (Table 3). Based on the results in Table 3 highly significant differences can be seen in the increments of the elements’ contents (p-value < 0.0001). Unlike the light soils, the years of research did not influence the differences in the quantitative changes of elements in various years (p-value = 0.357). In contrast, the interaction of the experimental factors was highly significant (p-value < 0.0001). Since all the nutrients differed significantly with respect to increments or losses, it was necessary to verify which of the contents of elements had similar tendency of quantitative changes under the influence of SSC application. For this purpose, homogeneous groups for increments or losses of element amounts were established (averages over the years of research). In Table 4, the results of the analysis show three distinct groups. The amounts of Corg (group a) presented the strongest increment and was a distinct group with N mineral and available S (group b). A separate group constituted available K (group c), for which amounts in soil were losses (increment ranging from 0 to 1). Other nutrients: P, Mg and Ntot (group ab) displayed a similar variability of contents (increments or losses).

Discussion

In the present study, two contrasting soils were used: light and medium, and it was hypothesised that during the pot experiment the rate and pattern of SSC decomposition will take place in a different way. Soil texture, physical properties and microbiological activity are the main factors governing the above process (Huang and Chen 2009; Järvan et al. 2017; Moretti et al. 2017). Coarse-textured soil similar to the light one is characterised by lower water-holding capacity and good aeration, whereas the conditions of fine-textured soil, like the medium one are the reverse. The physical properties are more favorable in light soil, and can accelerate the mineralisation process relatively soon after compost application. Regardless of this, the decomposition rate of compost organic matter, as well as native organic matter, should be quicker in light soil as opposed to medium soil. Moreover, in conditions of coarse-textured soils, a more obvious positive effect is expected of applied compost on soil’s chemical properties. Nevertheless these theoretical assumptions were not entirely confirmed by the study’s findings. The SSC showed a similar influence on both soils (Table 2). The application of SSC resulted in increased macronutrient amounts in soils, and this is well documented (Hargreaves et al. 2008; Bedada et al. 2016). The outcomes of this research do not confirm such information, because, independent of the soil agronomic category, the SSC had a weak influence on the amounts of nutrients in relation to the amounts obtained for the control soil (Figures 2 and 4, Table 2). This may result from the mature nature of the compost, which after composting process becomes a long lasting fertiliser, undergoing decomposition process slowly. However, the results obtained in this study (Figures 2–5, Table 2) agree with those of Graczyk and Jakubus (2016), Jakubus and Graczyk (2016), Giannakis et al. (2014), Duong et al. (2013), Huang and Chen (2009) who observed that SSC as an organic fertiliser undergoes slow mineralisation in soil. In this context, the importance of the composting process in stabilising organic compounds and reducing the proportion of soluble C, N, S and P forms should be stressed. From the agricultural point of view, soil organic carbon is a key property for defining the quality of soils under intensive management (Bonanomi et al. 2014). It has been demonstrated that the application of SSC statistically significant increases the amounts of soil organic carbon (Figures 2 and 4, Table 2). This relationship is clearly and thoroughly discussed in the literature (Castan et al. 2016; Hernandez et al. 2016; Jakubus and Graczyk 2016). Cited researchers reported that compost contain a significant portion of readily available organic carbon and in this connection can increase the pool of organic C in the soil.

Compost contains smaller amounts of the available N, P, S fraction as a consequence of the composting process, and thus it releases lesser amounts of these nutrients (Jakubus 2016; Rigby et al. 2016). The degree of compost stability also plays an important role in the balance of the mineralisation-immobilisation processes in amended soil. These transformations of nutrients were observed in the experiment presented here (Figures 3 and 5, Table 3). During the 3 year experiment, in particular years, the amounts of total and mineral N, as well as available P and S were subjected to different patterns of quantity changes expressed by both increments and losses. The lack of a clear effect of compost on the amounts of these nutrients can be explained by both the high background total N and available P and S in the soils and low mineralisation of compost. This is in accordance with the findings of this study, because a significant increment of Ntot was noticed in the soils after the 1st and 2nd years. After the last year of experiment, we found losses of Ntot in both soils, and this was strongly manifested in the case of the medium one. This tendency might be explained by the immobilisation process, which was also mentioned by Sevilla – Perea et al. (2014). Immobilisation of N, due to the increase of soil microbial biomass, is thought to be responsible for the observed low N availability in soils amended with sewage sludge compost and this immobilisation effect is greater in soil with a high clay content. It can be assumed that the addition of SSC not only increased immobilisation but also mineralisation of N originating from both compost and native soil organic matter. According to Rigby et al. (2016) the rate of N mineralisation in biosolid-amended soil is influenced by application rate, soil type and temperature, moisture content and pH value. In this study, there were differences between soils. After an initial decrease of Nmin amounts in medium soil (after the 1st and 2nd years), they subsequently increased (after the 3rd year) (Figure 5). The pattern of quantity changes in Nmin was different in the light soil because, after the 2nd year of experiment, an increment was noticed, followed by a loss in the soil after the 3rd year of experiment (Figure 3). These results support those of earlier studies. Sciubba et al. (2014) showed the tendency of N mineral to accumulate at low concentration percentages (10%). Giannakis et al. (2014) demonstrated that nitrogen availability is generally low and is estimated to range from 10% to 20% during the first year of application. In addition, Jakubus and Graczyk (2016) indicated low concentrations of labile N and a slow rate of compost decomposition, even with a higher dose of sewage sludge compost (120 Mg ha⁻¹).

The addition of SSC to soils should increase mineralisation of S. According to Zhao et al. (1996) the mineralisation rate of S in organic compounds mainly depends on their stability and organic S compounds added to soil usually decompose for more rapidly than native forms of soil organic S. However only in the case of light soil was it possible to observe increments of available sulphur (Figure 3). Soluble is potentially being replenished through organic matter decomposition. Taking into account the mature nature of compost used, the influence of native organic matter on quantitative changes of sulphur cannot be excluded.

Bedada et al. (2016) and Hargreaves et al. (2008) showed that amount of available P increased with the addition of compost to the soil. According to Hargreaves et al. (2008) a low mineralisation rate of P is seen immediately after compost application, but after a period of 3 months, compost provided sufficient P for plant growth. Giannakis et al. (2014) also found the improved status of available soil P as a result of compost application, although the increase was small and was not proportional to the application dose. The findings of this study (Figures 2 and 5, Tables 2 and 4) in terms of available P in soils are not comparable with the data in the literature. A significant increment of available P was noticed in medium soil after the 2nd year, but after the last year of experiment, there was a loss of this nutrient (Figure 5). In the light soil, significant losses of available amounts of P were observed after the 1st and 2nd years, but after the 3rd year a small increment was found (Figure 3). A possible explanation for this is related to the low decomposition of organic matter compost, which may have also delayed the release of available phosphorous. Jakubus (2016) showed that the release rate of P from composts strongly depends on the share of sewage sludge in the composting mixture. Higher bioavailable amounts of phosphorus were observed particularly in composts with small shares of sewage sludge (maximum 40%), characterised by rapid organic matter decomposition. The SSC applied in the present study was fine-textured and contained 50% of sewage sludge, thus when supplied to the soil, it caused a small increment of the amounts of available P in the light soil and a loss in the medium soil after the 3rd year of experiment (Figure 5, Table 2). The interactions of P in soil with organic matter and the metal cations such as Ca, Mg, Fe and Al should be underlined in the interpretation of the observed pattern of quantitative changes of P, because these complexes decrease the bioavailability of P (Fuentes et al. 2006; Castan et al. 2016). The conditions and properties of medium soil are more favorable to the occurrence of such a reaction resulting in the lowering of available P to plants.

Hargreaves et al. (2008) on the basis of the conducted literature review, stated that contents of K and Mg increased in soil even when very low rates of compost were used. The increases in K and Mg concentrations in the surface layer of soil amended with compost was noted by Bedada et al. (2016). Moreover Graczyk and Jakubus (2016) found that compost in higher doses (120 Mg ha⁻¹) had a significant impact on available amounts of potassium and magnesium in light soil. In this study, the compost dose was low – 6 Mg ha⁻¹ and the amounts of both Mg and K available to plants decreased in the soils during the 3 year experiment. It should be emphasised that SSC is not a valuable source of K and Mg. Likewise, the biomass provided by compost is likely the most influential factor contributing to the sorption process of potassium and magnesium cations. The sorption process together with plant uptake may explain the significant losses in available amounts of K and Mg both in the light and medium soils. This tendency was more strongly manifested in the case of potassium. The main factor in the loss of K in the investigated soils is related to the K fixation process, which is affected by higher content of clay minerals. Simonsson et al. (2009) confirmed that high clay content seems to be associated with large K fixation capacity. As a result of this, a greater decrease in available K amounts was observed in medium soil compared to the light one.

The cluster analysis used (hierarchical and non-hierarchical) made it possible to present a visualisation of the data and to show trends in the increments of groups of individual nutrients. The non-hierarchical k- means method, allowed for a visual confirmation of the well-known fact that the amounts of particular nutrients in medium soil are higher than those found in light soil. Meanwhile, the hierarchical method made it possible to notice trends in increments of nutrients for a particular type of soil fertilised with SSC. This facilitated the use of other standard statistical tools for data analysis, i.e. ANOVA and the Tukey method of homogenous subsets. However, these methods made it possible to distinguish nutrients differing significantly with respect to the increments of the contents of individual elements, or allowed us to create homogeneous groups. The statistical methods proposed and utilised (both cluster analysis and ANOVA) are the correct instruments for indicating if there are differences in the dynamic of SSC mineralisation and the rate of release of available forms of N, P, S, K and Mg in two contrasting soils. Thanks to the figures presented as dendrograms, the quantitative changes in the nutrients could be seen during the 3-year pot experiment and the trends in the changes could be identified which were common for particular nutrients independently of the soil analysed. The cluster analysis performed, and analysis of variance with respect to the average increments of various elements, as well as the post-hoc test (Tukey's test) demosntrated the slow decomposition process of SSC in both soils, and simultaneously confirmed that transformations of biogenic nutrients were based on several factors connected with soil properties as well as nutrient character.

To conclude, the fertilising effect of SSC was similar in both contrasting soils, although the conditions of the light soil were more favourable for quicker organic matter mineralisation and the release of available forms of N, S and P. SSC application significantly improved the status of organic carbon in the analysed soils. The weak influence on potential increments of available nutrient amounts should be related to the very slow decomposition process of SSC as well as to its application in a low dose (6 Mg ha⁻¹). The important role of compost organic matter and native organic matter in tendency of quantitative changes of biogenic nutrients should be taken under consideration in interpreting their behaviour in the environment. Considering the principles of sustainable agriculture, SSC seems an acceptable organic fertiliser, but, undoubtedly, in practise higher doses than in this study must be recommended in order to obtain a reasonable fertilising effect. Moreover, mineral fertilisation supplementing the available amounts of K and Mg must be obligatorily incorporated into management of the fertilisation system. The statistical methods applied have been shown to be a useful tool for visualisation and for data analysis in terms of practical evaluation of the fertilisation effect of SSC.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Monika Jakubus’ scientific career is tightly connected with Poznan University of Life Sciences. She is an Associate Professor at the Department of Soil Science and Ground Protection. Her main research areas of interest are: agricultural use of sewage sludge, compost and organic fertilizers; monitoring soils in terms of their contamination with heavy metals, sulphur and biogenic macronutrients; evaluation of nutrients availability for plants in various types of agricultural soils; sustainable agriculture and proper management of municipal organic wastes. She is the author and co-author of more than 100 papers published in Polish and international journals. Moreover, she was major advisor of doctoral and master thesis.

Ewa Bakinowska is an assistant professor at the Institute of Mathematics in Poznan University of Technology. She is working on statistical analysis of experimental data. The object of her scientific interest is in particular: generalized linear model, estimation methods in logistic models with multinomial distribution, log-linear model with Poisson distribution. Ewa Bakinowska analyzes the relationship between the estimation methods: geometrical methods (least squares method, weighted least squares method) and maximum likelihood methods (iterative methods: Newton-Raphson method and Fisher scoring method).

Bernard Gałka, employed in the Institute of Soil Science and Environmental Protection at the Wrocław University of Environmental and Life Sciences, is an expert in soil fertility assessment, soil contamination, and soil reclamation. His recent works focused on the impact of mountain forests reconstruction on the soil and habitat quality, and on the reclamation of soils polluted with heavy metals.