Determining the effect of exogenous organic materials on spatial distribution of maize yield

Knowledge on spatial distribution of crop yield in relation to fixed soil fertilisation with exogenous organic materials is essential for improving precise crop and soil management practices within a field. This study assessed the effect of various application rates and types of exogenous (recycled) organic materials (EOMs) containing different organic matter and nitrogen contents vs. mineral nitrogen on the yield of maize by means of linear regressions (trends), spatial kriging-interpolated maps, and Bland-Altman statistics. The experiments were conducted in 2013 and 2014 on two soils, i.e. loam silt in Braszowice (Poland) and clay silt loam in Pusté Jakartice (Czech Republic) under a cross-border cooperation project. The organic materials included compost from manure, slurry, and straw (Ag), industrial organic compost from sewage sludge (Ra), animal meal from animal by-products (Mb), and digestate from a biogas fries factory (Dg). The following 3 application rates of each EOM were adjusted according to the reference 100% = 200 kg N ha−1: 50 (50% N from EOM and 50% mineral N), 75 (75% N from EOM and 25% mineral N), and 100 (100% N from EOM). 100% mineral N was applied on control plots. All treatments were carried out in 4 replicates. The linear regressions between the EOM application rates and the maize yield were in general ascending in the Braszowice soil and descending in the more productive Pusté Jakartice soil. The spatial kriging-interpolated maps allowed separating zones of lower and higher yields with EOMs compared to the control. They were attributed in part to the different EOM application rates and soil water contents. The Bland-Altaman statistics showed that addition of 50% of N from EOMs in 2013 caused a decrease and an increase in the maize grain yield in Braszowice and Pusté Jakartice, respectively, whereas the inverse was true with the 75 and 100% EOM additions. In 2014, the yield of maize for silage increased with the increasing EOM application rate in Braszowice and decreased in Pusté Jakartice, but it was smaller on all EOM-amended plots than in the control. As shown by the limits of agreement lines, the maize yields were more even in Pusté Jakartice than Braszowice. These results provide helpful information for selection of the most yield-producing EOM rates depending on the site soil conditions and prevalent weather conditions.

Soil organic matter (SOM) is a key factor affecting many ecosystem services including nutrient cycling that support crop production worldwide [1][2][3][4] . However, its content in soil decreases due to land-use change from natural forest or perennial grasslands to cultivated croplands; then, it is frequently subjected to intensive soil disturbance by conventional tillage practices 3,5,6 . It was reported that about 42 to 78 Gt of carbon were lost globally 7 due to tillage and narrow crop rotation without legumes and cover crops and enhanced biological decay and soil erosion 8,9 . The loss of SOM from the topsoil of 20 cm due to soil disturbance and inadequate return of organic matter within a 30-50-year time period has been assessed as high as 50% 10 . The rate of SOM loss is increasing with progressive warming related to climate change 3,11 . The total rate of biomass and SOM losses is estimated to be up to 20% of worldwide atmospheric carbon dioxide emissions 12,13 . Therefore, the decline in SOM is considered as an environmental threat and a soil degradation component, as specified in the Communication from the European Commission "Thematic Strategy for Soil Protection" 14 .
To avoid further reductions and reverse the current SOM content, an international research 4 per mille initiative has been launched at the Conference of the Parties (COP21) in Paris meeting (2015), which requests to increase carbon stocks at 4 per 1000 (or 0.4%) per year on agricultural lands to mitigate the SOM losses caused by climate change and to increase food production for the rising world population while being environmentally

Materials and Methods
A field experiment was carried out on loamy silt soil in Braszowice, Poland (50° 34′ 03.0″N 16° 48′ 07.4″E), and clay silt loam in Pusté Jakartice, Czech Republic (49° 58′ 23.5″N 17° 57′ 19.8″E), in 2013 and 2014 under a cross-border cooperation project 25,26 . The soils contain 14 and 20% of clay and 1.02 and 1.13% of organic carbon, respectively. Both field sites are located about 140 km apart from each other. The experimental field in each site received 10 treatments, i.e. 3 exogenous organic materials (EOMs) × 3 rates and the control, with digestate from a biogas fries factory (Dg) in Braszowice, compost from manure, slurry, and straw (Ag) in Pusté Jakartice, and industrial organic compost from sewage sludge (Ra) and animal meal from animal by-products (Mb) in both fields. The following 3 application rates of each EOM were adjusted according to the reference 100% = 200 kg N ha −1 : 50 (50% N from EOM and 50% mineral N), 75 (75% N from EOM and 25% mineral N), and 100 (100% N from EOM). 100% mineral N was applied on the control plots. The treatments were replicated in 4 blocks (A, B, C, D), which gave forty plots. Each plot had a surface area of 20 m 2 in Braszowice and 25 m 2 in Pusté Jakartice. The layout of the experments is shown in Fig. 1. The total nitrogen contents were 6.9, 2.6, 2.3, and 8.4% in Dg, Ag, Ra, and Mb, respectively. The corresponding organic carbon contents were 40.7, 24.1, 17.9, and 40.1%. The Mb, Ra, and Ag were applied twice in Pusté Jakartice. Ra and Dg were applied twice and Mb once in Braszowice. Instead of phosphorus-containing Mb, a mineral N fertiliser was applied in the second study year in Braszowice to avoid the possible risk of eutrophication. A conventional tillage system including moldboard ploughing to a 25-cm depth was applied in both sites in late autumn, followed by a cultivator (15 cm) and pre-sowing surface preparation with a harrow in spring to prepare the seedbed. The EOMs were applied on the soil surface and then mixed into a 15-cm depth by a cultivator before seedbed preparation and sowing maize (Zea mays L.). The variety N K Terada, FAO 260 was grown in 2013 for grain and Ulan-FAO 270 was cultivated in 2014 for silage in both sites. On each plot, maize was grown in four rows and six rows in Braszowice and Pusté Jakartice, respectively. The maize was harvested by hand cutting at a height of 10 cm from two central rows to determine the grain and straw yield in 2013 and the silage yield in 2014 27 .
Basic statistics including the average, standard deviation, coefficient of variation, minimum and maximum values, skewness, kurtosis, and linear regressions between the EOM application rates and the yield components were calculated. Spatial dependence and distribution of the yield components were evaluated using geostatistical methods. Experimental semivariograms were computed and then mathematical functions were adjusted to semivariograms that were used for mapping by ordinary kriging 28,29 .
In order to determine a separate effect of the different application rates of N from the EOMs vs. control plots (only mineral nitrogen fertiliser) on the maize yield, the Bland-Altman statistics was adopted 22,30 . In this graphical method, the differences in the maize yields between the plots with different EOM nitrogen application rates and the control plots against the average yield with and without EOMs were determined. The agreement between the yield in the EOM-amended and control plots was assessed using bias (average of differences between the yields from the EOM-amended and control plots), the limit of agreement (LoA) defined as bias ± 1.96 × standard deviation, confidence intervals (CI) for the bias and LoA defined as ± standard error × the value of t distribution with n-1 degrees of freedom, and the Bland-Altman ratio (BAR) defined as the ratio of half the range of LoA to the mean of the pair including the yield from the EOM-amended and mineral nitrogen fertilized plots. The agreements were graded as good, moderate, and insufficient for BAR values < 0.1, 0.1-0.2, and >0.2, respectively 22 . Root mean square residuals (RMSR) and maximum relative residuals (MRR), which are the difference in the yield between the EOM-amended and control plots were determined for all yield components in 2013 (grain, straw, and grain plus straw) and 2014 (silage yield).
Maize yield. The means and the minimum and maximum values of all yield components in both study years, except for straw, in 2013 were greater in Pusté Jakartice than Braszowice ( Table 2). The grain yield in Pusté Jakartice in 2013 was higher by 37.5% than in Braszowice (10.07 t ha −1 ), while the straw yield was higher by 15.8% in Braszowice than in Pusté Jakartice (11.34 t ha −1 ). The yield of maize for silage in 2014 was higher in Pusté Jakartice by 16.7% than in Braszowice (16.52 t ha −1 ). Irrespective of the yield component and study year, the CV values ranged from 12.4 to 17.1% in Braszowice and from 7.5 to 9.7% in Pusté Jakartice. The asymmetry (skewness) of the yield components was positive in Braszowice (0.087-0.294), whereas it varied in 2013 from positive 0.162 for the straw yield to negative −0.076 for the grain yield in Pusté Jakartice. The kurtosis ranged from positive 0.418 to negative −0.186 in Braszowice and from positive 0.126 to negative −0.874 in Pusté Jakartice. The skewness and kurtosis values indicate that the yield components were close to the normal distribution.
The effect of the increasing EOM application rate on the maize yield was related to the study site, year, and yield component (Fig. 3). In Braszowice, the yield of both the maize grain and the straw yield displayed a general ascending trend with the increasing EOM application rate. The most pronounced trend was noted for the combined yield of grain and straw in 2013 and the least distinct trend was found for the silage yield in 2014 when the yield remained almost the same at all EOM types and application rates. In Pusté Jakartice, however, the trends were in general descending, irrespective of the study year and maize yield component.

Bland-A ltman plots.
The linear regressions discussed in the section above demonstrate a foremost tendency to change the maize yields in response to the EOM addition. The use of Blant-Altman plots, however, allowed determining the impact of the particular application rates of EOMs on the maize yield. Figures 4 and 5 show horizontal lines of bias, limits of agreement (LoA) along with confidence intervals (CI), and regression lines intersecting or overlapping with bias lines. In general, the values of CI for bias were relatively small and considerably greater with LoAs.
As indicated by the bias values (<0), the application of 50% N from EOMs in Braszowice in 2013 resulted in lower grain, straw, and grain plus straw yields compared to the control (only mineral N) (Fig. 4). Maize yield  www.nature.com/scientificreports www.nature.com/scientificreports/ components increased with the increasing application rates from 75 to 100% N from EOMs (bias >0) and were greater compared to the control. In 2014, as indicated by the bias (<0) lines, the yield of maize for silage increased with the increasing EOM application rate but was smaller on all EOM-amended plots than in the control. In Pusté Jakartice, however, the addition of only 50% N from EOMs increased the grain yield in 2013 (bias >0) in contrast to the other EOM application rates (Fig. 5). The straw and grain plus straw yields decreased with the increasing EOM application rate (bias <0) and were lower than in the control plots. In 2014, the yield of silage maize was lower on the EOM-amended than control plots.
Comparison of Figs. 4 and 5 indicates that the range of limits of agreement (LoA) was from 1.5 to more than 3 times greater for Braszowice than Pusté Jakarice, indicating that the maize yield was more even in the latter. This was confirmed by the similarly larger values of root mean square residuals (RMSR) and maximum relative residuals (MRR) for Braszowice than Pusté Jakartice ( Table 3). The Bland Altman ratio value (BAR) indicates that the LoA values were very good or satisfactory (according to 22 ) for Pusté Jakartice (0.72-0.153) and in general moderate for Braszowice (0.139-0.236). Hence, the results from Braszowice are burdened with greater uncertainty in 2013 but are much more certain in 2014. The Bland-Altman plots indicate that, regardless of the experimental site and study year, a majority of the points are within the limits of agreement (LoA) and those outlying -within the confidence intervals (CI) (Figs. 4 and 5). The regression line of the yield differences between the EOM-amended and control plots against the average yield of both indicates that the trend in 2013 was only descending in Pusté Jakartice, whereasit was descending, ascending, or almost unchanged (close to bias line) in Braszowice. In 2014, however, the trend was ascending regardless of the EOM application rate and study site. The intersections of the regression lines with the bias lines can be indicative of the yield-producing effectivenes of EOMs. The values of the yield differences above the bias lines, generally in the regions to and from the intersection point of the regression line with the bias line, indicate a greater yield-producing effect of nitrogen from EOMs than from the mineral fertiliser. However, the values below the bias lines indicate a lesser effect of EOMs on the yield than mineral fertilisation. The greater or lesser effect of EOMs on the yield of maize may be related to the type of EOM. This aspect requires further research taking into account the availability and uptake of nitrogen by plants from individual EOMs.
Semivariogram models and parameters. The semivariogram spherical models were fitted to the empirical semivariograms for the maize yield (Table 4)   www.nature.com/scientificreports www.nature.com/scientificreports/ Maps of maize yield. As can be seen in Fig. 6, two distinct zones can be separated with similar yield responses in both sites corresponding approximately to half the field length. These zones are from 0 to 45 m and 45 to 90 m in Braszowice and from 0 to 30 m and 30 to 60 m in Pusté Jakartice. Such separation can be related to the random distribution of plots fertilised with different types and application rates of EOMs, which was the same in both sites (Fig. 1). It is worth noting that the concentration of plots with the large doses of EOMs in some  www.nature.com/scientificreports www.nature.com/scientificreports/ a larger yield was noted in the area from 18-30 m in the lower part in Pusté Jakartice, where plots with a dose of 50% N from EOMs dominated (Fig. 1). Mapping of maize yield can support variable application of nitrogen from EOMs across the field.

Discussion
The regression lines (trends) showed that the maize yield tended to increase and decrease with the increasing EOM application rate in Braszowice and Pusté Jakartice, respectively although the absolute yield was greater in the latter (Fig. 2). This tendency was less pronounced in Pusté Jakartice. The greater and less variable absolute maize yield in Pusté Jakartice than Braszowice can be related to the greater soil water content at most occasions, as reported in the accompanying paper 31 and thereby better water supply for maize. These differences in the soil water content can be attributed in part to the greater water holding capacity in Pusté Jakartice (33.82 m 3 m -3 ) than Braszowice (31.80 m 3 m -3 ) 26 as a result of better soil aggregation and the presence of a greater number of earthworms that we observed during the field measurements. Both stabilize the pore structure and permit storage of greater amounts of water in soil profile. Futhermore, more variable maize yield in the less productive Braszowice than Pusté Jakartice soil can in part result from its greater sensitivity to changes in soil water content associated with EOM application.
The study allowed identifying the specific benefits of three statistical approaches for description of maize yield response to different levels of exogenous organic matterials. Classical statistics allowed to recognize the general trends of the impact of exogenous organic matterials on yield based on mean values, dispersion, variability and asymmetries in distribution and kurtosis. However, the use of geostatistical methods including semivariograms and kriging-interpolated maps permitted delineating low-yielding sub-field areas on two soils of different agricultural suitability under variable application of oxogenous organic matters. The Bland-Altman plot analysis showed for the first time in what manner the partial and total replacement of readily available nitrogen from the synthetic fertiliser with N from exogenous organic materials affected the maize yield in particular sites. In Braszowice, a positive increase in the grain yield occurred only after the application of the larger doses of organic N (75 and 100% N from EOMs). In turn, in the Pusté Jakartice soil, this was observed already after the addition of the smallest dose, i.e. 50% N from EOMs. The less sensitive grain yield response to organic N in Braszowice vs. Pusté Jakartice can be elucidated by the visually observed less friable structure in the former that diminishes the ability of roots to grow and uptake water and nutrients 33 . The more efficient yield-producing effect of organic N in Braszowice than Pusté Jakartice was also observed with the yield of silage maize. The different maize yield    Table 3. Root mean square residuals (RMSR), maximum relative residuals (MRR), and Bland-Altman ratio (BAR) of differences in the maize yield between the EOM-amended and control plots. www.nature.com/scientificreports www.nature.com/scientificreports/ response to the EOM application in Braszowice and Pusté Jakartice revealed by the Bland-Altman statistics is in line with the spatial yield distribution on the kriging maps. The maps showed that, in areas with dominance of plots amended with 100% N from the EOMs, the grain yield was larger in Braszowice and smaller in Pusté Jakartice compared to the neighboring areas with the majority of plots amended with 50% N from EOMs. These results are of practical importance, as they indicate that more organic N is required in the less productive soil with poorer structure in Braszowice than Pusté Jakartice to increase the maize grain yield. They will help to develop management strategy to reduce losses of the unused by crops nitrogen through volatilization and leaching into the environment as well as farm input costs depending on environmental conditions. Recent studies (e.g. 33,34 ) have shown that nitrogen uptake and utilisation efficiency in maize can be influenced by tillage systems.
Review of the literature indicates that the effect of soil amendment with organic matter should be considered not only with respect to crop production but together with the broader ecosystem services and benefits, including carbon sequestration 7,18,35 , climate change mitigation and adaptation 3,9 , improvement of resistance to soil erosion and soil compaction 3,36,37 , and better natural pest control 4 , which lead to reduced reliance on mineral N fertilisers or other farm chemicals and lower environmental pollution 4,38 . Moreover, biological research on the same fields as those used in this study showed that the addition of EOMs stimulated the activity of dehydrogenases and the poorly explored microbial functional and genetic diversity in soil 39 . It also demonstrated that EOMs are not a threat to the soil edaphone 40 and concentration of carbon dioxide in the atmosphere 41,42 . Given the yield response observed in the present study and the ecosystem benefits, precise application of the exogenous organic materials offers an opportunity for sustainable crop production and environmental protection as well as proper management of organic wastes from agricultural and biogas production systems.
The Bland-Altman plots also demonstrate that the regression lines of maize yield in 2013 were close to the bias lines in both sites, while they increased with the increasing average yield from the EOM-amended and control plots in 2014. These interannual differences may be related to the relatively high sum of precipitations in 2013 during intensive maize growth (May-June) than in the other period and during the later growth (July-September) www.nature.com/scientificreports www.nature.com/scientificreports/ in 2014 in both places. This difference may have caused diversification of the spatial distribution of maize yields through different spatial storage and redistribution of water depending on the EOM application rate. Furthermore, the range of the limits of agreement (±1.96 SD lines) of the Bland-Altman plots for both the grain and straw yield is about two times higher for Braszowice than Pusté Jakartice in both study years. This proves that the soil in Pusté Jakartice creates more even growth conditions for plants than that in Braszowice. However, the soil in Braszowice requires optimised and localised treatments to improve and smooth plant growth conditions. Further studies are intended to assess the most yield-producing EOM rates under range of soil and weather conditions and types of exogenous materials using the statistical methodology verified in this work.

Summary and Conclusions
The yield of maize in response to the application of four exogenous (recycled) organic materials (EOMs) at three application rates was explored and assessed by means of linear regressions (trends), spatial kriging-interpolated maps, and Bland-Altman statistics in two experiments. The regression trends of maize yield in general increased with the increasing EOM application rate in Braszowice and decreased in the more productive Pusté Jakartice soil. The spatial kriging-interpolated maps allowed delineating zones of lower and higher maize yields that were related to the spatial distribution of the EOM application rates. The concentrations of plots with the large doses of EOMs were in general reflected in the increased yield in Braszowice and the reduced yield in Pusté Jakartice. The analysis of the Bland-Altaman statistics revealed that, in 2013, the effect of 50% N from EOMs on the maize yield was negative and positive in Braszowice and Pusté Jakartice, respectively, whereas the inverse was true for the additions of 75 and 100% N from EOMs. However, in 2014 characterised by a greater sum of precipitations during the growing season, the increasing EOM application rate caused an increase and decrease in the maize silage yield compared to the control plots in Braszowice and Pusté Jakartice, respectively. The limits of agreement showed that the maize yield was more variable in Braszowice than in Pusté Jakartice. The study confirmed the hypothesis that the use of the Bland-Altman method allows determining separate effects of various EOMs application rates on the maize yield compared to the maize yield grown on plots fertilised only with mineral nitrogen. The results will support precise adjusting the most yield-producing nitrogen rates from EOMs depending on the study site conditions and weather conditions prevailing during growing season.