To assess the agronomic effects of the digestate (DG) in comparison with undigested cattle slurry (CS) or with mineral fertilisation (MF) a pot experiment was carried out with annual ryegrass (Lolium multiflorum Lam., cv. Winter Star II) as plant test. The pots were placed outdoor under a glass cover (39˚49'25.9"N 7˚27'07.5"W) from 29^th November 2018 to 19^th May 2019 at the Polytechnic Institute of Castelo Branco, Portugal. During the crop cycle the monthly means of air temperature ranged from 8.4˚C to 18.4˚C. The soil used was a dystric regosol [19] developed on granitic rock in a Mediterranean climate. It was an acid (pH_H2O = 5.1) sandy loam soil (8% clay, 18% silt and 74% sand), with medium cation exchange capacity (12.5 cmol_c·kg⁻¹) compared with reference values (CEC 10.1 - 20.0 cmol_c·kg⁻¹) of the method (ammonium acetate buffered at pH 7.0; [20]), low levels of exchangeable Ca, Mg, and Na (2.0, 0.67 and 0.09 cmol_c·kg⁻¹ respectively). It had a medium content of organic matter (27 g·kg⁻¹). The soil had also a low content of available P (8 mg·kg⁻¹of Olsen P), medium level of exchangeable K (0.3 cmol_c·kg⁻¹) and a low degree of base saturation (20%). In the Mediterranean region these soils are mainly used for rain-fed pastures and for cereal crops such as rye and barley.

The soil was sampled at 0 - 0.20 m depth, air dried and sieved through a 5 mm mesh and poured into plastic pots each one with 6 kg of capacity. The experiment was carried out during 170 days. Since the soil is poor in available P it was done treatments with and without mineral P fertilisation. The P fertilisation was done with single superphosphate (SSP) at a rate of 35 kg P ha⁻¹ incorporated into the soil at the beginning of the experiment before sowing. The mineral nitrogen fertilisation was done with a total amount of 170 kg N ha⁻¹ the threshold level to soil N application from organic amendments by the Portuguese legislation [21]. So, the mineral N fertilisation was divided into three applications: 85 kg N ha⁻¹ at sowing and then 42.5 kg N ha⁻¹ as top-dressing after the first cut of the ryegrass and the remaining 42.5 kg N ha⁻¹ after the second cut. Concerning the N fertilisation through the organic amendments it was done by two different ways: 1) DG or CS were mixed with the soil one week before sowing using an amount corresponding to the full application of 170 kg N ha⁻¹ (No), in this case no more N was applied during the crop cycle and 2) DG or CS were mixed with the soil one week before sowing using an amount corresponding to the application of 85 kg N ha⁻¹ and then mineral N was applied at a rate of 42.5 kg N ha⁻¹ as top-dressing after the first cut of the ryegrass and the remaining 42.5 kg N ha⁻¹ after the second cut (NoNi). The mineral N was applied to soil in solution using ammonium nitrate (NH₄NO₃) reagent p.a. The experiment consisted of a randomized block design with ten treatments each one with four replicates generating a total of 40 pots. The treatments used are referred in Table 1.

After DG and CS addition to soil at 29^th November 2018, the soil was watered at 70% of the field capacity. After one week the ryegrass was sown with 60 seeds per pot. The cuts of the ryegrass were done at 71, 111 and 163 days after sowing. The pots were watered throughout the crop cycle at 70% of the field capacity.

After each cut of ryegrass, dry-matter production was quantified and the concentration of N and P in dry matter was determined. Also, a soil sample of each pot was taken after the end of the experiment for soil analysis.

The agronomic effects of the organic amendments were assessed by 1) soil fertility properties, 2) crop yield (dry matter production) and 3) indexes of N and P-use efficiency of the crop.

The indexes of N and P-use efficiency of the crop [22] [23] used were 1) N or P uptake, 2) apparent N or P recovery by crop (Equation (1)) and 3) agronomic efficiency of N or P (Equation (2)).

Apparent N or P recovery by crop ( % ) = [ ( U n − U c ) / F n ] × 100 (1)

In Equation (1) U is the uptake (mg N or P kg⁻¹ soil) with (Un) or without (Uc) N or P fertilisation and Fn is the amount of N or P applied (mg N or P kg⁻¹ soil).

Agronomic efficiency , AE ( gDMYmg − 1   P ) = [ ( Y n – Y c ) / F n ] (2)

In Equation (2) Y is the dry matter yield (g DMY kg⁻¹soil) with (Yn) or without (Yc) N or P fertilisation and Fn is the amount of N or P applied (mg N or P kg⁻¹ soil).

In order to get a better understanding about the effects of the organic amendments on P availability the P recovery (Prec) by the Olsen method [24] (Equation (3)) was also calculated using the following ratio (units in mg P kg⁻¹ soil):

Prec ( % ) = [ ( Olsen P after the experimente − Initial Olsen P ) / ( Padded to soil ) ] × 100 (3)

Soil

Soil samples were air dried, sieved (<2 mm) and analysed for textural class, organic matter, pH (H₂O), electrical conductivity (EC), available P, and exchangeable bases (Ca, Mg, K and Na). The soil texture was evaluated by the particle size analysis using the pipet method [25]. The organic matter was analysed according to the procedure described in [26]. The pH was measured using a pH electrode in a 1:2.5 soil to solution ratio and the electrical conductivity with a conductivity meter in a 1:2 soil to water ratio suspension. The contents of the exchangeable bases Ca, Mg, K and Na were measured by atomic absorption spectrometer after extraction with a molar solution of CH₃COONH₄ buffered at pH 7.0. The Olsen P was used to evaluate plant available P, which was determined according to Olsen et al. [24] and it was quantified by the method of Murphy and Riley [27].

Ryegrass

Shoots of rye grass of each treatment and cut were weighed for quantification of the biomass production (fresh matter) then they were dried at 65˚C for 48 h and weighed again for the quantification of crop yield on a dry matter production basis. The dried plants were grounded in a ball mill, sieved at a 0.5 mm-mesh sieve and then were placed in a muffle furnace at a temperature of 480˚C for 16 h to obtain the ashes. Total P was measured after digestion of the ashes with hydrochloric acid solution (HCl 20%, v/v). P was quantified by spectrophotometry at a wavelength of 470 nm. Nitrogen was quantified by the Kjeldahl procedure (N_k) after each cut and before drying the shoots.

Organic amendments

The organic amendments, solid digestate (DG) and undigested cattle slurry (CS) were characterised for moisture content by the gravimetric method; pH was determined with a glass electrode (organic amendment to water ratio of 1:5) and electrical conductivity with a conductivity meter in a 1:10 suspension (organic amendment to water ratio). Organic matter was quantified by the loss of weigh after ignition at 550˚C during 16 h. Total N was evaluated by Kjeldahl procedure. Total P was quantified after digestion with aqua regia solution (using 3.0 g DM + 7.5 ml HNO₃ 65% + 21 ml HCl 37% during 2 h at 120˚C [28]) and then quantified by molecular absorption spectrophotometry. Total cations such as Fe, Mn, Cu, Pb, Cd, Ni and Cr were also extracted by the aqua regia digestion and quantified by atomic absorption spectrophotometry [28]. Total K and Na were quantified by flame emission spectrophotometry and total Ca and Mg by atomic absorption spectrophotometry. For the quantification of total K, Ca, Mg and Na it was used an hydrochloric acid solution (HCl 37% diluted in H₂O at a ratio of 1:1) of the ashes obtained after the ignition at 550˚C of the organic amendments (ashes from 3 g DM + 75 ml of the hydrochloric acid solution).

Statistical analysis of the data was carried out with IBM SPSS statistics 26 software (https://www.ibm.com/products/spss-statistics). One-way ANOVA analyses were conducted to identify the agronomic effects of the treatments through the analysed soil properties, the crop yield (dry mater production) and the indexes of N and P use efficiency by the crop. Tukey’s test was used to compare means at 0.05 probability level.

The chemical composition of the solid digestate (Table 2) showed that it is a product in the solid state although with a high moisture content (710 g·kg⁻¹). This value of moisture is within the range (800 - 570 g·kg⁻¹) refereed to the non-dried solid fraction of digestates (after a liquid-solid separation) by Möller [29] and Teglia et al. [30]. Consequently, the digestate had a low amount of dry matter (290 g·kg⁻¹) which had almost 56% of organic matter. Although a large proportion of the DM of our digestate is organic its value is lower than that reported for Tambone et al. [7] which ranged between 67% - 74% for digestates obtained from livestock slurries.

DG is referred as an alkaline product with a pH value in the range 7.3 - 9.0 [31] [32]. Our digestate had also a pH within that range with a value of 7.8 and had high amounts of Ca (31.3 kg·Mg⁻¹ in the Fresh Matter-FM) and also considerable amounts of Mg (4.5 kg·Mg⁻¹ FM) compared with values for other digestates like 1.0 - 2.3 kg·Mg⁻¹ FM for Ca and 0.3 - 0.7 kg·Mg⁻¹ FM for Mg [33] [34] [35]. In relation to the K concentration DG showed an amount of 4.6 kg·Mg⁻¹ FM within the range (1.2 - 11.5 kg·Mg⁻¹ FM) reported by Möller et al. [36]. Since K is not a structural nutrient it occurs in dissolved forms either in the plant cells

¹All the parameters analysed in the digestate are expressed on a dry matter basis excluding density, pH and CE.

and in the digestates therefore, the liquid phase of the digestates is characterised by higher K contents than the solid phase [12]. DG had low values of Na (0.9 kg·Mg⁻¹ FM) and also of electrical conductivity (EC of 1.7 dS·m⁻¹). EC of digestate was within the normal range considering that 3 dS·m⁻¹ is the threshold value for irrigation water quality to prevent soil salinity [37]. Although most of the total P is allocated to the solid phase our DG has a low amount of P (0.72 g·kg⁻¹ DM) compared with the values referred by other authors 2 - 35 g·kg⁻¹ DM [6] [34] [36] [38]. In general, the liquid phase of the digestates had a higher proportion of mineral N than the solid fraction [36]. Despite our DG has been pressed after digestion it had a total N content of 27.3 g·kg⁻¹ DM within the values indicated by Tegila et al. [30] for digestates obtained from animal slurries without phase separation but lower than the values (total N of 53 - 151 g·kg⁻¹ DM) reported by Tambone et al. [7]. Nevertheless, 85% of the total N in DG was in organic forms, and this fact together with a C:N ratio of 12 suggests also that the rate of N mineralisation of DG may be slow and may not release N at a rate similar to that of crop needs for the time of a single crop cycle. In turn, CS is a fluid material which showed a low amount of dry matter (55.8 g·L⁻¹) and of organic matter (44.2 g·L⁻¹). It had a pH in the neutral range (pH of 6.9) lower than DG. Pötsch et al. [38] referred an increase of about 0.8 units in the pH value of digestates compared with undigested livestock slurries. That increase in pH value of DG affects the equilibrium between the inorganic-N forms of NH₃/NH₄ and favours the increase of N losses by volatilisation of N-NH₃ in DG. CS had lower amounts of Ca and Mg (2.4 and 0.6 g·L⁻¹ respectively) and higher amounts of K and Na (3.0 and 1.1 g·L⁻¹) than DG. CS had a high level of electrical conductivity (13.8 dS·m⁻¹) which limits the annual volume to be used in agricultural soils in order to prevent soil salinity. CS had a low content of P (0.018 g·L⁻¹) and a considerable content of total N (2.7 g·L⁻¹). It had 56% of total N in organic forms and a lower C:N ratio (C:N of 9) which suggests that CS can have a faster mineralisation rate than DG and thus, in principle, could provide N to the crop needs soon after its incorporation into the soil.

The levels of other nutrients in DG and CS such as Cu, Zn Fe, Mn and Ni and non-nutrients such as Cd, Cr and Pb (data not shown) were below the threshold levels compared with those established for agricultural use of composts by the Portuguese legislation [39]. Concerning the microbiological analysis, the digestate was free of microorganisms of faecal origin such as Escherichia coli which had 6.3 × 10² CFU·g⁻¹ and Salmonella spp. which was not detected. These values are also in accordance with the above-mentioned legislation [39].

The standards of some European countries [39] [40] [41] specify for products to be considered as soil amendments (those that have a beneficial effect on soil OM) that their OM content should be between 20% - 90% in a DM basis and should have a dry matter content higher than 30% - 50% [39] [42]. In relation to the C/N ratio it should be higher than 8 [42]. So, overall the digestate meet the standards except for the dry matter content slightly below the minimum threshold level (29%).

Soil fertility

While the digestate and the undigested cattle slurry presented contrasting OM contents, the content of soil organic matter after the experiment (Figure 1(a)) had low variations and did not change significantly (P > 0.05) between the treatments. The soil OM ranged between 28.3 g·kg⁻¹ in NoPi_DG and 33.2 g·kg⁻¹ in NoNi_CS. Indeed, although the digestate had 558 g·kg⁻¹ of OM on a dry matter basis it has only 162 g·kg⁻¹ of OM in the fresh matter. So, the amount of OM incorporated into soil in the No_DG treatment was only 3.5 Mg ha⁻¹ and in the No_CS treatment was 2.8 Mg ha⁻¹. The relatively low values of the C:N ratios of the digestate and the cattle slurry (12 and 9 respectively) together with the relatively low amount of OM incorporate into soil didn’t favour an increase of the OM of the soil through a single application. Teglia et al. [30] referred that values higher than 40 for the OM/N_org ratio implies a low organic nitrogen content of the digestates and thus their beneficial effects as soil amendments could be limited due to a nitrogen immobilisation after soil incorporation. The OM/N_org of our DG is 24, it has a C/N ratio of 12 higher than 8 considered the threshold level by the French criteria for soil amendments [42] and its C/N ratio is also in the range of 10 - 20 a criterion for the stability and maturity of soil amendments [43] [44]. So, the lack of response to increase soil organic matter will be due mainly to the low amount of digestate incorporated into soil as a consequence of its high moisture content (710 g·kg⁻¹).

Concerning the EC values after the experiment the fertilisation with N and P leads in all the treatments to a significantly increase (P < 0.001) in the soil EC (Figure 1(b)) compared with the control treatment and with treatments without P application. However, all treatments had very low EC values (0.04 in N0P0 and 0.1 dS·m⁻¹ in NoPi_DG) far from 0.4 dS·m⁻¹ considered the soil threshold level to avoid saline effects for the crops. Although the fertilisation led to significant variations in the soil pH (Figure 1(c), P < 0.001) the soil remains acidic with values ranged between 5.1 (NiPi) and 5.6 (No_DG and No_CS). The control treatment showed a pH value of 5.4. The significantly higher pH in treatments with the application of digestate or cattle slurry as the only fertiliser suggests that the content of basic cations (namely Ca and Mg) of these organic amendments should have a significant effect on the decrease of soil acidity and could promote better conditions for plant nutrition in the acidic soils. It was observed also that the sum of the soil exchangeable bases (Figure 2) increased significantly (P < 0.001), namely the exchangeable Ca, in the treatments with the digestate application (No_DG).

Although the digestate had considerable amounts of Ca and Mg a single application of digestate was not enough to increase the soil fertility class of the exchangeable bases which remained at low levels (Ca @ 2.5, Mg @ 0.86, K @ 0.16 and Na @ 0.1 cmol_c·kg⁻¹). The treatments fertilised with superphosphate (Figure 3(a)) showed a significantly increase (P < 0.001) compared with the control treatment in soil available P, measured by the Olsen method.

However, this increase is more evident in the treatments fertilised with superphosphate together with digestate. The treatment NoPi_DG has the highest Olsen-P level (26 mg·kg⁻¹) and the NoNi_CS treatment had the lowest value (10 mg·kg⁻¹). In addition, DG showed also a significantly increase (P < 0.001) in the P recovery (87% and 64% in NoPi_DG and NoNiPi_DG, respectively) by the Olsen method (Equation (3) and Figure 3(b)) compared with the NiPi treatment (50%) and even with CS for similar treatments (54% and 33%). In other work with the same type of soil Horta [45] observed a reduction in soil sorption capacity after a compost addition to soil accompanied by an increase in soil available P. This P behaviour was also observed in other works [46] [47] [48] and could be explain by the competition with the phosphate anion for adsorption sites by the organic anions of the organic material (the digestate in our experiment) produced during its decomposition into the soil.

Ryegrass yield

Overall the fertilisation significantly increased (P < 0.001) the ryegrass yields irrespective of the nutrients source (Table 3).

The control treatment (N0P0) showed the lowest cumulative ryegrass yield (10.03 g DM pot⁻¹) and the highest yield was obtained in the treatments with only mineral fertilisation NiPi (33.68 g DM pot⁻¹) or in DG and CS treatments but with half of the N fertilisation from those organic sources and the other half from mineral N NoNiPi_DG (28.33 g DM pot⁻¹) and NoNiPi_CS (29.67 g DM pot⁻¹) but always with mineral P fertilisation. When the total amount of N (170 kg·ha⁻¹) was applied only by DG or CS the yields decreased significantly (P < 0.001) in relation to treatments with any N mineral fertilisation.

Different letters in the columns indicate for each cut and for the cumulative dry matter yield significant differences by the Tukey test at the P < 0.05 level.

Concerning the comparison between DG treatments with and without P fertilisation the yields didn’t change significantly but for CS the NoPi_Cs and No_Cs were significantly different. These results suggest that the main shortcoming for ryegrass nutrition with DG should be the N availability but for CS should be not only N but also P availability. In addition, the ryegrass yields between the three cuts showed a great variability (Table 3) caused, in principle, by changes in the N availability of DG and CS throughout the crop cycle. Thus, in the first cut the DG and CS treatments with NoPi showed higher yields than the NiPi treatment (P < 0.001). In turn in the second cut the NiPi treatment showed the highest ryegrass yield (P < 0.001) and only NoNiPi_CS had a similar yield. Finally, at the end of the crop cycle the highest yield was obtained only in the NiPi treatment. These results suggest that throughout the crop cycle the availability of N from DG or CS changed. Soon after DG or CS application in treatments with only No application the fertilising effect is better than NiPi in principle because of a higher N availability than NiPi (in NiPi treatment the Ni application rate at sowing was only 85 kg·ha⁻¹).

Indeed, in the DG and CS treatments with NoNi there were no significant differences from NiPi indicating a similar N availability. But in the second and third cut the availability of N form DG or CS was not enough for the ryegrass needs. In the second cut the lower N availability of DG compared with CS could be explained by a slower mineralisation rate and even a partial N immobilisation of the inorganic N applied as topdressing after the first cut. Indeed, other authors [49] [50] observed also a partial N immobilisation after the addition to soil of a solid digestate from cattle slurry and justify it by an increase in the soil microbial activity caused by the addition of easily degradable organic matter from DG.

Nitrogen and phosphorus use efficiency indexes

Concerning the efficiency indexes, we can observe that the N uptake (Figure 4(a)) was higher (P < 0.001) in the NiPi treatment but without significant differences from the DG treatments NoNi and of NoPi_CS and N0NiPi_CS. Although in the first cut the uptake of N in No_DG treatments were similar to NiPi at the end of the crop cycle the total N uptake was similar to the control treatment. This behaviour confirms the decrease in N availability from DG in the second and third cuts. Regarding P uptake (Figure 4(b)) it was higher in DG treatments (except in No_DG) and in CS treatments with Pi fertilisation (NoPi_CS and NoNiPi_CS) in comparison with NiPi. This higher P uptake in treatments fertilised with organic amendments compared with only mineral fertilisation may be due to an improvement in soil P availability (Olsen P and P recovery increased in DG treatments, Figure 3) caused by the role of the organic matter on the decrease of soil P sorption of the applied P and/or an increase in soil P desorption as explained above. However, the behaviour of CS in improving soil P availability could be lower than that of DG as observed by the deeper decrease in P uptake in treatments without P application in CS compared with DG. Liedl et al. [15] observed an increase in soil available P with solid digestate application over a four-year field experiment. Nevertheless, their digestate had a higher amount of P than that we used in our work, in fact we only applied 4 kg P ha⁻¹ from DG and 1 kg P ha⁻¹ from CS in the NoPi treatments. Such low P addition from the organic amendments is not enough to justify by itself the observed increases in P uptake.

The other indexes like the N agronomic efficiency and the N apparent recovery (Equations 1 and 2, Figure 5) showed significant differences between the treatments (P < 0.001) with similar behaviour between them as referred above for N uptake. In relation to these indexes the N apparent recovery is higher in NiPi, NoNiPi and NoNi with DG and CS while the N agronomic efficiency was higher in NiPi and only in NoNiPi treatments highlighting the need of complementary Ni and Pi fertilisation to obtain a N and P efficiency use similar to NiPi treatment. Möller and Müller [12] referred that DG applied directly with incorporation into soil immediately after field spreading provide plant available NH₄ content plus a small part of the No fraction (10% - 20%). In our work DG had 85% of total N in organic forms so, the amount of available N seems to be enough in the short-term at least for 50% of all the ryegrass needs in N.

At the end of the crop cycle the P apparent recovery ranged between 27% in NiPi and 43% in NoNiPi_DG (P < 0.05, Equation (1), Figure 6(b)) with no significant differences between the other treatments. The P agronomic efficiency was significantly higher (P < 0.01) in NiPi, NoNiPi_DG and NoNiPi_CS treatments (Equation (2), Figure 6(a)). The results of the agronomic indexes seem to

confirm that there is a shortcoming on N uptake by the ryegrass in the treatments with only No fertilisation (No_DG and No_CS) caused by a decrease in N availability, in the short-term, which caused also a decrease of the P uptake and an overall decrease of the agronomic indexes.