Double-annual forage crop rotation controls nutrient surpluses in N based organic fertilization

The use of organic fertilizers from stock-raising activity is usually based on nitrogen (N) criterion. The objective of our research is to evaluate whether this N criterion lead to positive or environmentally risky changes, mainly in organic carbon (OC) storage and the availability of the main nutrients. Three biennial rotations of four crops were set up (in a 6–year time-frame). The treatments included a control (no N added), two mineral treatments where 250 kg N ha -1 yr -1 were provided at two different times during the rotation, three cattle manure treatments which provided 170, 250 and 500 kg N ha -1 yr -1 and four treatments in which the two lowest manure rates were complemented with mineral N (80 and 160 kg N ha -1 yr -1 ). Over the whole 6-year period, the measured soil OC increases were equivalent to ca. 25, 43 and 35% of the manure OC applied respectively, following the manure rate increases. Furthermore, equivalences were of -5, 23 and 25% when compared with full mineral fertilization. The positive slope of phosphorus availability was 1.2 mg Olsen–P kg -1 to 10 kg P ha -1 applied from manures (N:P=4) when starting from a threshold of 15.9 mg Olsen–P kg -1 . The availability of other nutrients (Mg, Zn) and Na also increased with rates. In a medium-term experiment, the manure rate of 250 kg N ha -1 yr -1 optimized the nutrient recirculation. In the long– term, rotations should be redesigned to control P surpluses or the amount of N applied from manures should be reduced.


INTRODUCTION
implemented, with two annual rotations: oats (Avena sativa L.) -sorghum (Sorghum bicolor L.) in the first year, and ryegrass (Lolium multiflorum L.) -maize (Zea mays L.) in the second year. The experiment continued for three cycles of the two-year rotation (a 6-year period), beginning in October 2007. Winter crops (oats and ryegrass) were maintained in the field from October-November to May-June, while summer crops (sorghum and maize) were maintained from May-June to October-November. Time between the harvest of the preceding crop and the sowing of the next one varied from two to three weeks. All crops were used as forage. The fertilization experiment was designed as a randomized complete block with 10 treatments (based on N criterion) and 3 replications.
The amount of manure to be applied was determined by its analytical composition (Table 2) and according to N criterion. The treatments (Table 3) included a control (0-0) where no N was applied, two treatments with 250 kg of N mineral ha -1 yr -1 applied at the sowing time of winter and summer crops (250M-0) or applied only as topdressing (0-250M). Three treatments included cattle (Bos taurus) manure equivalent to 170,250 and 500 kg N ha -1 yr -1 (170C-0, 250C-0 and 500C-0, respectively). The lowest rate was applied at the sowing of the summer crop only. For the other two manure rates, applications were divided between the two crops (summer and winter). The two lowest manure rates (170 and 250 kg N ha -1 yr -1 ) were complemented with mineral N at rates of 80 (170C-80M or 250C-80M) or 160 kg N ha -1 yr -1 (170C-160M or 250C-160M), which added 4 new treatments. Solid manures came from a neighboring farm, with the exception of the 2008 summer application where a more liquid animal waste was used.
During the whole experimental period, fertilizing treatments were continued in the same plots. Plot size was 50 m 2 (5 m x 10 m). of 24h after their application. Mineral N fertilizer was applied as ammonium nitrate sulfate (26% N) at sowing, and as urea (46% N) at topdressing, following the usual fertilization practices in the area. Phosphorus and potassium fertilization were planned to avoid P and K nutrient deficiencies. Crop demand was based on biomass production ( Fig. 2) and the analyzed P and K concentrations or from references (Table 4). Models of crop yield evolution according to N treatments have been already published (Perramon et al., 2016). Phosphorus concentration was quantified in the planted crops during the first rotation and was analyzed using ultraviolet spectroscopy (UV). The K concentration was measured in oat and sorghum during the first year and was analyzed by atomic emission spectroscopy (AES). Ryegrass and maize K data were obtained from the Agricultural Department of the region (Generalitat de Catalunya, 2014); although K concentrations could vary across the range of fertility treatments. When no manure was applied or at the low manure rate, P and K were added annually as complex mineral fertilizer (0-14-14) and as potassium sulfate (50% K 2 O) in order to reach rates close to 130 kg P 2 O 5 ha -1 yr -1 and 260 kg K 2 O ha -1 yr -1 in each plot (Table 3). From the third year, the theoretical amount of K to be applied reached 380 K 2 O ha -1 yr -1 as we detected in the first rotation a theoretical K deficit of 31 kg K ha -1 (Table S1).
Oxidizable organic matter was determined by dichromate oxidation and subsequent titration with ferrous ammonium sulphate (Yeomans and Bremner 1998). Oxidizable C stock (kg OC ha -1 ) at 0-0.3 m depth was calculated using the measured bulk density.
The amount of sequestered organic C (kg OC ha -1 ) in 0-0.3 m soil depth was estimated after subtracting the oxidizable C stock in the control (0-0) or in the mineral treatment (250M-0) to other treatments. Soil bulk density was evaluated by the core method (Soil Survey Staff, 2011) in 250M-0, 170C-0, 250C-0 and 500C-0.

Statistical analysis
Different analyses of variance (ANOVA) were performed by the GLM procedure, which is included in the minimum quadratic methods for the adjustment of linear models (Table S2). When the year or the rotation was omitted (Table S3)

RESULTS AND DISCUSSION
At the end of the experimental period, no significant differences in soil bulk densities were found between selected treatments (Table S3), despite decrease in bulk density as the amount of OM applied increased (Table 7). The average value obtained (1565 kg m -3 ) is consistent with the soil texture (Balba, 1995) and was used for soil nutrient calculations.
Analysis of variance of different soil chemical proprieties indicated significant changes in some of them (Table S3 and Tables 5 and 6) depending upon the treatment in the 0-0.3 m depth. No differences in OM, N, P and K contents were found in the deeper layers (Table 6).
Only in two cases over a total of twelve cattle manure applications, a maximum C:N ratio of 24.6 was recorded, the rest being lower than 20. The average ratio of 14 (Table   2) indicates the predominance of a net N mineralization of the organic N applied (Probert et al., 2005). In the short-term, N release (C:N< 13) or N immobilization (C:N >15) could coexist in a rotation according to the C:N of the manure applied (Qian and Schoenau, 2002) and depending upon other factors such as water availability (Kim et al., 2011). When applied to soils (C:N=7.6, Table 1), the soil OM content increased significantly at the two highest manure rates (250C-0, 500C-0) or when the lowest rate was complemented with mineral N (170C-80M, 170C-160M; Table 6).
The organic matter increment when using manure has been well described by different authors (Grignani et al., 2007;Monaco et al., 2008;Tomasoni et al., 2011). It also contributes to C storage, which means that the carbon concentration increased (compared with 0-0) by 0.25, 0.43 and 0.35 kg C kg -1 OC applied for 170C-0, 250C-0 and 500C-0, respectively (Fig. 3). The first figure was close to 23% found by Bhogal et al. (2009). The higher storage capacity of our soils is probably explained by their initial lower content (8.7 g OC kg -1 , Table 1) when compared with those (11-33 g OC kg -1 ) from Bhogal et al. (2009). When C storage was accounted for by comparing it with a non-limited N scenario (250M-0, mineral) differences were -0.05, 0.23 and 0.25 kg C kg -1 OC applied for 170C-0, 250C-0 and 500C-0, respectively. This means that the EU permitted N rate (170C-0), without a mineral N complement, constrains yields ( Fig. 2) and potential C storage (Fig. 3). Treatments 170C-80M and 170C-160M showed a linear trend to increase total Kjeldahl N ( Fig. 4a) and soil OC stock (Table 6) when mineral N is added.
The equivalent to 86% of the N applied with manures remained in the soil at the end of the six-year period which surpasses the proportion of 72% of N applied as organic-N ( Table 2). The reduction of a need for the mineral N complement to reach maximum yields at the two lowest manure rates (Fig. 2), indicates higher N availability with time (from re-mineralization of newly formed soil OM). This positive effect of the yearly manure incorporation (Fig. 2) makes statistically significant the interaction rotation*treatment (Table S2) in the yield analysis. This N residual effect from manures already described by various authors (Schröder, 2007;Muñoz et al., 2008) is close to 80 kg N ha -1 in our experiment which agrees with Schröder's et al. (2005) figures.
Nevertheless, the EU allowed rate (170C-0) still requires a minimum complement of 80 kg N ha -1 to maximize yields and a bit more to enhance C storage (Fig. 3). In fact, in a similar system, Grignani et al. (2007) found an optimum productive rate which agrees with our 170C-80M treatment. Furthermore, they also pointed out an N deficit for the control (124 kg N ha -1 yr -1 ) which was close to our deficit average of 99 kg N ha -1 yr -1 and as a result, in both cases, N deficits constrained yields and enhanced mining of nutrients.
Phosphorus surplus is a common trend when using manures (Nyiraneza et al., 2009;Moore et al., 2014). According to Sileshi et al. (2017), our manure ratio C:P< 200 (ca. Table 2) indicates the predominance of net P mineralization over P immobilization.

52,
Besides, its N:P ratio average of 4 (Table 2) can be associated with P accumulation in soil (Table 6) when manures cover N demand (250C-0 and 500C-0), as the rate of the crop extractions (determined for the 4 crops in the first two years) was 4.6 (the whole above-ground plant matter is used for forage). Although there was an absence of significant differences in P concentrations at the lowest manure rates after the 6 year experimental period (Table 6), a positive linear P accumulation trend ( Fig. 4b) was evident. This linear function (from a threshold of 22.38 mg kg -1 ) agrees with Leytem et al. (2011) working in soils with a pH range of 7.5-7.9. However, our slope was half that of these authors, probably because their experiment lasted just 5 weeks, just covering the period of high increase of phosphatase production by soil microbes. Phosphorus added from manure and other sources tends to become less available to plants with passing time (Wandruszka, 2006). In our intensive forage system, from 16 mg kg -1 of available soil P (Olsen), available P stock roughly increased at a rate of 5.4 kg ha -1 (0-0.3m) for 10 kg of P applied with manure ( Fig. 4b), up to a concentration of 80 mg P kg -1 (500C-0). Nevertheless, to sustain the option of maximum N recycled (250C-0, 250 kg N ha -1 ) the introduction of other crops (e.g. legumes) in the rotation could be of interest, because the attained Olsen P values higher than 30 mg kg -1 are considered very high for a sandy loam soil (Rodríguez Martín et al., 2009). The maximum accumulation recorded in treatment 500C-0 is a severe environmental threat. As suggested by Heckrath et al. (1995) on a more fine-textured soil (clay loam), once the concentration of Olsen-P exceeds a certain concentration (60 mg P kg -1 in his experiment) an enhanced contribution of P losses through subsurface runoff appears. These P losses will increase the potential for water eutrophication (Sims et al., 1998), although in our experiment, no significant differences in P content were found in deeper horizons (Table 6).
In the third rotation, the yield increases reduced theoretical K surpluses down to 8.5 and 2.5 kg K ha -1 yr -1 in 170C-80M and 250C-80M treatments, respectively (Table S1).
There was no deficit because the increase in crop demand was compensated with a higher average concentration of K in the applied manure in the last two years (Table 2).
Soil K level rose in 500C-0 achieving a high concentration (Fig. 4c), but without rising its content at deeper layers. It was maintained in acceptable ranges (Cottenie, 1980) in the rest of the treatments. The concentrations of the two exchangeable cations Mg and Na, and the one from Zn increased following the manure rate increases, being significant versus the mineral fertilization in 250C-0 and 500C-0 treatments. The Zn increment agree with other authors (Nyiraneza et al., 2009) and the accumulation would require a long-term study to evaluate potential negative impacts as described by Benke et al. (2008).

CONCLUSIONS
In intensive and highly productive double annual forage crop rotations, the equivalent manure rate of 250 kg N ha -1 yr -1 maximizes yields and OC storage (0-0.30 m depth).
The C, N, P and K distribution in the deeper layers (0.30-0.90 m) is not affected.
However, in the long term, the observed tendency to accumulate P in the 0-0.30 m soil layer could be environmentally risky. Rotations should be redesigned or else manure addition should be stopped for a short period of time to help in the control of P surpluses. If the manure rate is reduced to the European permitted rate in nitrate vulnerable areas, equivalent to 170 kg N ha -1 yr -1 for N of organic origin, C storage and the recirculation of nutrients will be limited. Besides, an additional mineral complement of N (80 kg N ha -1 yr -1 ) and K (to attain 380 kg K 2 O ha -1 yr -1 ) will be required to satisfy crop demand and to maintain soil fertility levels. The results of this study provide guidelines for farmers, in similar agricultural systems, regarding the choice of an appropriate N application regime to maximize the circular nutrient economy of their farms.

TABLES
First ( Table 3. Description † of mineral (M) and cattle manure (C) fertilization treatments (kg ha -1 ) in a two year rotation (four crops) based on N manure rates including organic (Org-N) and inorganic (Inorg-N) nitrogen. Nitrogen, P, and K complementary rates ‡ by mineral fertilizer to cover crop demand are included.
Treatments.  0 1178 0 † Information before the hyphen indicates the fertilizer treatment at sowing and after the hyphen the treatment at cereal tillering. Numbers before the letter C indicate the amount of N applied (kg N ha -1 ) from cattle manure at sowing. The rate of 170 kg N ha -1 (170C) was only applied to the summer crop (sorghum or maize). When a higher rate was applied, it was divided into 100 and 150 kg N ha -1 (250C) or into 250 and 250 kg N ha -1 (500C) for the winter and summer crop, respectively. The numbers before the code M indicate the amount of N applied (kg N ha -1 ) from mineral fertilizers. When 80M or 160M were applied as a topdressing, they were fractioned into 30 or 60 kg N ha -1 for the winter crop and into 50 or 100 kg N ha -1 for the summer crop, respectively. When only mineral N was applied (250M), it was divided into two applications of 100 (winter crop) and 150 (summer crop) kg N ha -1 both applied at sowing (250M-0) or as a topdressing (0-250M) in both crops. At sowing, ammonium nitrosulfate (26%) was applied and urea (46%) was applied in topdressings. ‡ When 170 kg N ha -1 was applied as cattle manure, P and K rates were complemented using a N-P-K mineral fertilizer (0-14-14) and potassium sulfate (50% K 2 O). They were applied to cover the maximum crop demand. § In the first rotation, any fertilizer was applied to the summer crop (sorghum) as topdressing. . Table 4. Potassium (K), phosphorus (P) and nitrogen (N) concentrations (g kg -1 ) for the different crops included in the first rotation and used to calculate theoretical crop demand.
Treatment † Oat Sorghum Ryegrass Maize K ‡ P ‡ N K ‡ P ‡ N ‡ K § P ‡ N ‡ K § P ‡ N ‡ 0-0 11.8 1.6 8.1 20.5 2.2 6.7 11.3 1.4 8.8 8.8 2.2 7.1 250M-0 14.2 1.6 8.6 14.2 1.7 8.9 11.3 1.4 6.9 8.8 1.8 9.5 0-250M 12.2 1.8 9.1 18.4 1.9 -11.3 2.3 13.1 8.8 1.9 12.4 170C-0 11.4 1.5 9.1 18.2 1.9 6.8 11.3 1.5 10.2 8.8 2.5 7.8 170C-80M 11.4 1.5 8.2 18.9 1.6 -11.3 1.9 9.6 8.8 2.3 9.7 170C-160M 12.8 1.9 9.9 18.7 1.7 -11.3 2.3 12.7 8.8 2.3 10.3 250C-0 11.4 2.1 8.9 17.9 1.8 6.8 11.3 1.5 8.1 8.8 2.8 8.5 250C-80M 11.0 2.0 9.0 17.8 2.0 -11.3 1.8 8.5 8.8 2.3 9.5 250C-160M 13.4 1.7 9.6 16.5 2.3 -11.3 2.1 10.8 8.8 2.2 11.3 500C-0 11.5 1.7 8.2 18.4 2.2 8.3 11.3 1.6 7.4 8.8 2.7 8.7 † Information before the hyphen indicates the fertilizer treatment at sowing and after the hyphen the treatment at cereal tillering. Numbers before the letter C indicate the amount of N applied (kg N ha -1 ) from cattle manure at sowing. The rate of 170 kg N ha -1 (170C) was only applied to the summer crop (sorghum or maize). When a higher rate was applied, it was divided into 100 and 150 kg N ha -1 (250C) or into 250 and 250 kg N ha -1 (500C) for the winter and summer crop, respectively. The numbers before the code M indicate the amount of N applied (kg N ha -1 ) from mineral fertilizers. When 80M or 160M were applied as a topdressing, they were fractioned into 30 or 60 kg N ha -1 for the winter crop and into 50 or 100 kg N ha -1 for the summer crop, respectively. When only mineral N was applied (250M), it was divided into two applications of 100 (winter crop) and 150 (summer crop) kg N ha -1 both applied at sowing (250M-0) or as a topdressing (0-250M) in both crops. At sowing, ammonium nitrosulfate (26%) was applied and urea (46%) was applied in topdressings. When 170 kg N ha -1 was applied as cattle manure, P and K rates were complemented using a N-P-K mineral fertilizer (0-14-14) and potassium sulfate (50% K 2 O). They were applied to cover the maximum crop demand. In the first rotation, any fertilizer was applied to the summer crop (sorghum) as topdressing. ‡The concentrations are obtained from the analysis. §The concentrations followed Generalitat de Catalunya (2014). Table 5. Soil chemical fertility † (0-0.3 m depth) after 6 years of the experiment establishment. † Information before the hyphen indicates the fertilizer treatment at sowing and after the hyphen the treatment at cereal tillering. Numbers before the letter C indicate the amount of N applied (kg N ha -1 ) from cattle manure at sowing. The rate of 170 kg N ha -1 (170C) was only applied to the summer crop (sorghum or maize). When a higher rate was applied, it was divided into 100 and 150 kg N ha -1 (250C) or into 250 and 250 kg N ha -1 (500C) for the winter and summer crop, respectively. The numbers before the code M indicate the amount of N applied (kg N ha -1 ) from mineral fertilizers. When 80M or 160M were applied as a topdressing, they were fractioned into 30 or 60 kg N ha -1 for the winter crop and into 50 or 100 kg N ha -1 for the summer crop, respectively. When only mineral N was applied (250M), it was divided into two applications of 100 (winter crop) and 150 (summer crop) kg N ha -1 both applied at sowing (250M-0) or as a topdressing (0-250M) in both crops. At sowing, ammonium nitrosulfate (26%) was applied and urea (46%) was applied in topdressings. When 170 kg N ha -1 was applied as cattle manure, P and K rates were complemented using a N-P-K mineral fertilizer (0-14-14) and potassium sulfate (50% K 2 O). They were applied to cover the maximum crop demand. In the first rotation, any fertilizer was applied to the summer crop (sorghum) as topdressing. ‡ NS: non-significant (p > 0.05). Column averages with a different letter indicate significant differences (α = 0.05) based on Duncan's multiple range test.
Treatment † pH Electrical Cation Ca Mg Na Cu Zn Mn conductivity exchange capacity ---------(cmol c kg -1 ) -----------(DTPA, mg kg -1 ) ----(1:2.5) (1:5; dS m -1 ) (cmol c kg -1 ) 0-0 8.0 0.14 10.6 9.6 0.73e 0.04c  Table 6. Soil organic matter, total N, available P and K soil content at different depths (0-0.30, 0.30-0.60, 0.60-0.90 m) after 6 years of the experiment establishment. † Information before the hyphen indicates the fertilizer treatment at sowing and after the hyphen the treatment at cereal tillering. Numbers before the letter C indicate the amount of N applied (kg N ha -1 ) from cattle manure at sowing. The rate of 170 kg N ha -1 (170C) was only applied to the summer crop (sorghum or maize). When a higher rate was applied, it was divided into 100 and 150 kg N ha -1 (250C) or into 250 and 250 kg N ha -1 (500C) for the winter and summer crop, respectively. The numbers before the code M indicate the amount of N applied (kg N ha -1 ) from mineral fertilizers. When 80M or 160M were applied as a topdressing, they were fractioned into 30 or 60 kg N ha -1 for the winter crop and into 50 or 100 kg N ha -1 for the summer crop, respectively. When only mineral N was applied (250M), it was divided into two applications of 100 (winter crop) and 150 (summer crop) kg N ha -1 both applied at sowing (250M-0) or as a topdressing (0-250M) in both crops. At sowing, ammonium nitrosulfate (26%) was applied and urea (46%) was applied in topdressings. When 170 kg N ha -1 was applied as cattle manure, P and K rates were complemented using a N-P-K mineral fertilizer (0-14-14) and potassium sulfate (50% K 2 O). They were applied to cover the maximum crop demand. In the first rotation, any fertilizer was applied to the summer crop (sorghum) as topdressing. ‡ NS: non-significant (p > 0.05). Column averages with a different letter indicate significant differences (α = 0.05) based on Duncan's multiple range test.
Treatment † Organic matter (g kg -1 ) Total N(Kjeldahl, g kg -1 ) P (Olsen, mg kg -1 ) K ( NH 4 AcO 1N, mg kg -1 )  Table S1. Estimated nutrient (P, K) balances (nutrient applied -nutrient removed; kg ha -1 ) from the different rotations. † Information before the hyphen indicates the fertilizer treatment at sowing and after the hyphen the treatment at cereal tillering. Numbers before the letter C indicate the amount of N applied (kg N ha -1 ) from cattle manure at sowing. The rate of 170 kg N ha -1 (170C) was only applied to the summer crop (sorghum or maize). When the rate of 250 kg N ha -1 (250C) was applied, it was divided into 100 and 150 kg N ha -1 for the winter and summer crop respectively. The numbers before the code M indicate the amount of N applied (kg N ha -1 ) from mineral fertilizers. When 80M or 160M were applied as a topdressing, they were fractioned into 30 or 60 kg N ha -1 for the winter crop and into 50 or 100 kg N ha -1 for the summer crop, respectively. When only mineral N was applied (250M), it was divided into two applications of 100 (winter crop) and 150 (summer crop) kg N ha -1 both applied at sowing (250M-0) or as a topdressing (0-250M) in both crops. At sowing, ammonium nitrosulfate (26%) was applied and urea (46%) was applied in topdressings. When 170 kg N ha -1 was applied as cattle manure, P and K rates were complemented using a N-P-K mineral fertilizer (0-14-14) and potassium sulfate (50% K 2 O). They were applied to cover the maximum crop demand. In the first rotation, any fertilizer was applied to the summer crop (sorghum) as topdressing.  Table S2. Analysis of yield biomass (kg ha -1 ) variance for the different fertilization treatments.