The Inﬂuence of the Application Technique and Amount of Liquid Starter Fertilizer on Corn Yield

: The aim of this research was to study the impact of application technique and rate of liquid starter fertilizer applied with a novel device on the production of corn. Starter fertilizer was applied in the root system range of freshly germinated plants in the ‘belt’ and ‘point’ forms at di ﬀ erent quantities (35, 50, 70, and 100 L ha − 1 ), which led to intensive plant growth in the initial stages of development. This adapted system was used for sowing and for application of the liquid starter fertilizer at the same time. The ﬁeld trial was set up at two sites (two di ﬀ erent land types), in the conditions of the natural water regime of the soil during the three vegetation seasons in the period 2016–2018. For this purpose, a prototype of the electronic device EUKU-01 was designed. The starter fertilizer was applied at 5 cm laterally from the row where the sowing was performed and 5 cm below the depth at which the corn seeds were sown. Data were statistically analyzed by two-factor analysis of variance, where the inﬂuence of mineral fertilizer treatment and the inﬂuence of liquid starter fertilizer treatment were observed as factors. The results showed that the optimal choice of the technique of liquid starter fertilizer application can result in fertilizer savings by 30% without reducing yield. fertilizer applied with a novel device on the production of corn was presented in this study. It shows that the yield results were obtained by applying mineral fertilizers (T 0 , T 1 and T 2 ) and di ﬀ erent application techniques and amounts of starter fertilizer (TT 0 , TT 1 , TT 2 , TT 3 and TT 4 ). Starter fertilizer was applied in the root system range of freshly germinated plants in the ‘belt’ and ‘point’ forms at di ﬀ erent quantities (35, 50, 70, and 100 L ha − 1 according to the two-by-two placement method, which led to in initial stages in were analyzed by two-factor analysis where liquid starter 30% the ‘point’ form, approximately case a


Introduction
Over the last two decades of the 20th century and the first decade of the 21st century, the average corn yield in the world has increased by 70%. This increase has been the result of constant progress in the breeding and development of increasingly fertile hybrids, application of different types and forms of fertilizers, but also the development of agricultural machines that are used to perform the necessary technological operations [1][2][3].
Sowing of corn, as a technological phase of production, is one of the most important elements of production technology because it directly affects the achieved yield [4,5]. Shortcomings and irregularities made during sowing can hardly be corrected by other cultural measures, which directly leads to a reduction in yield [6].
The trials were set up on soil types mentioned above. Soil samples were taken at depths of 0-30 cm and 30-60 cm before basic tillage for each year of study. Methods for chemical analysis of soil are used [32][33][34][35][36][37][38][39][40][41] and the results are shown in Table 1. During the research in the experimental variants ( Figure 1), conventional mineral fertilizers NPK (N:P:K = 15:15:15) and KAN with 27% N (13.5% NH 4 -N and 13.5% NO 3 -N) were applied in the following amounts: 0 + 0 kg ha −1 ; 150 + 100 kg ha −1 and 300 + 200 kg ha −1 . Based on available research [30,31,42,43], the amount of 100 L ha −1 of starting fertilizer was taken as the optimal value. Also, some of the research in the field [38] indicates that, in case of 'belt' form application, corn plants effectively adopt only 70% of the total amount of applied starter fertilizer. Therefore, in this research it was assumed that the same effect will be achieved if 'belt' form amounts were replaced with 30% less than the amounts in 'point' form. In addition to the above, liquid starter fertilizer Starter-A (N:P:K 10:40:10, inorganic, liquid) was applied in the amounts of: 100 L ha −1 'belt' form (i.e., 30% less is 70 L ha −1 in 'point' form) and 50 L ha −1 'belt' form (i.e., 30% less is 35 L ha −1 in 'point' form). The chemical composition of the applied liquid starter fertilizer is shown in Table 2.

Trial Design and Applied Corn Production Technology
Field trials were carried out using the split-plot method, from experimental plots E1 to E15 (see Figure 1). The size of all plots on which different treatments were applied was 11.2 × 90 m. Each of the plots, from E1 to E15 contains 16 rows. In all three years of research, winter wheat was a preceding crop to corn at the aforementioned sites. After the wheat harvest, the plant remains were removed and the stubble was plowed down to a depth of 10-15 cm. In the fall, before plowing (30 cm depth), mineral fertilizers were spread in different amounts. Mineral fertilizer was not applied to the plots from E1 to E5. Moreover, 150 kg ha −1 was applied to the plots from E5 to E10, and 300 kg ha −1 of NPK fertilizers were applied to the plots E10 to E15. In the spring, immediately before the preparation of the land for sowing, the mineral fertilizer KAN was applied in the amount of 100 kg ha −1 to plots E5 to E10 and 200 kg ha −1 to plots E10 to E15. After the application of mineral fertilizers, pre-sowing preparation was performed on two swaths using a corn planter and a spike harrow.
Sowing of medium-early corn hybrid ZP 427 (Maize research Institute "Zemun Polje", Belgrade, Republic of Serbia) was performed using an IMT634.454 (Industry of Machinery and Tractors, Belgrade, Republic of Serbia) corn planter with a row spacing of 70 cm and 65,000 ha −1 plants. Sowing was done in the second decade of April for all three years of research using the IMT-634.454 corn planter on all plots from E1 to E15. No starter fertilizer was applied to plots E5, E10 and E15 (Control plot). The adapted corn planter was used on the other plots where the liquid starter fertilizer (Starter-A) was applied. Adapting the corn planter and installing the prototype of the electronic device EUKU-01 enabled automatically sub-surface applying of the set doses of liquid starter fertilizer at the same time when sowing corn.
Within the conducted research, using the prototype of the electronic device EUKU-01, two technical systems of automatic subsurface application of liquid starter fertilizer were applied. The fertilizer was applied to the 'point' and 'belt' forms ( Figure 2). The application of liquid starter fertilizer in the belt form meant the application of fertilizer in the form of a continuous belt 25 mm wide next to the row in which the corn seeds were sown. The second method of application meant that the liquid starter fertilizer was applied to the point form (diameter r = 25 mm) individually next to each sown seed. In both cases, the liquid starter fertilizer was applied laterally from the row in In all three years of research, winter wheat was a preceding crop to corn at the aforementioned sites. After the wheat harvest, the plant remains were removed and the stubble was plowed down to a depth of 10-15 cm. In the fall, before plowing (30 cm depth), mineral fertilizers were spread in different amounts. Mineral fertilizer was not applied to the plots from E1 to E5. Moreover, 150 kg ha −1 was applied to the plots from E5 to E10, and 300 kg ha −1 of NPK fertilizers were applied to the plots E10 to E15. In the spring, immediately before the preparation of the land for sowing, the mineral fertilizer KAN was applied in the amount of 100 kg ha −1 to plots E5 to E10 and 200 kg ha −1 to plots E10 to E15. After the application of mineral fertilizers, pre-sowing preparation was performed on two swaths using a corn planter and a spike harrow.
Sowing of medium-early corn hybrid ZP 427 (Maize research Institute "Zemun Polje", Belgrade, Republic of Serbia) was performed using an IMT634.454 (Industry of Machinery and Tractors, Belgrade, Republic of Serbia) corn planter with a row spacing of 70 cm and 65,000 ha −1 plants. Sowing was done in the second decade of April for all three years of research using the IMT-634.454 corn planter on all plots from E1 to E15. No starter fertilizer was applied to plots E5, E10 and E15 (Control plot). The adapted corn planter was used on the other plots where the liquid starter fertilizer (Starter-A) was applied. Adapting the corn planter and installing the prototype of the electronic device EUKU-01 enabled automatically sub-surface applying of the set doses of liquid starter fertilizer at the same time when sowing corn.
Within the conducted research, using the prototype of the electronic device EUKU-01, two technical systems of automatic subsurface application of liquid starter fertilizer were applied. The fertilizer was Agriculture 2020, 10, 347 5 of 13 applied to the 'point' and 'belt' forms ( Figure 2). The application of liquid starter fertilizer in the belt form meant the application of fertilizer in the form of a continuous belt 25 mm wide next to the row in which the corn seeds were sown. The second method of application meant that the liquid starter fertilizer was applied to the point form (diameter r = 25 mm) individually next to each sown seed. In both cases, the liquid starter fertilizer was applied laterally from the row in which the corn seed was sown, at a distance of 5 cm, as well as below the depth of 5 cm at which the corn seed was sown. Two-by-two placement of starter fertilizer was used by Gatiboni [44]. For weed control, a combination of the herbicides Laudis (2 L ha −1 ) and Callisto (200 g ha −1 ) was applied to all plots (E1 to E15) in all three years of research at both sites. A tractor in aggregate with an AGS 440 (Agromehanika, Kranj, Republic of Slovenia) sprayer was used for the application of herbicides. Interrow cultivation was performed using an IMT-626.40 (Industry of Machinery and Tractors, Belgrade, Republic of Serbia) interrow cultivator.
The yield of pure corn grain was obtained by separately harvesting each variant and reducing it to 14% grain moisture. Harvesting was performed in the optimal time, with a universal grain harvester CLAAS LEXION 430 (CLASS Group, Harsewinkel, Germany), and before calculating the yield for each trial variant, the water content in the corn grain was determined by drying at 105 °C. Grain yield with the water content of 14% was calculated by the Equation (1):

The Working Principle of the Prototype of the Electronic Device EUKU-01
Schematic and basic construction components of the prototype of the electronic device EUKU-01 (Ciga Factory, Aranđelovac, Republic of Serbia) is shown in Figure 3.  For weed control, a combination of the herbicides Laudis (2 L ha −1 ) and Callisto (200 g ha −1 ) was applied to all plots (E1 to E15) in all three years of research at both sites. A tractor in aggregate with an AGS 440 (Agromehanika, Kranj, Republic of Slovenia) sprayer was used for the application of herbicides. Interrow cultivation was performed using an IMT-626.40 (Industry of Machinery and Tractors, Belgrade, Republic of Serbia) interrow cultivator.
The yield of pure corn grain was obtained by separately harvesting each variant and reducing it to 14% grain moisture. Harvesting was performed in the optimal time, with a universal grain harvester CLAAS LEXION 430 (CLASS Group, Harsewinkel, Germany), and before calculating the yield for each trial variant, the water content in the corn grain was determined by drying at 105 • C. Grain yield with the water content of 14% was calculated by the Equation (1):

The Working Principle of the Prototype of the Electronic Device EUKU-01
Schematic and basic construction components of the prototype of the electronic device EUKU-01 (Ciga Factory, Arandelovac, Republic of Serbia) is shown in Figure 3.
During sowing, the sowing apparatus of the corn planter performs individual sowing of corn seeds which after separating from the sowing plate fall into the open sowing furrow. Photoelectric sensors (position 2) placed under the sowing devices; at the moment of falling (movement of the grain from the sowing plate to the sowing furrow) generate signals which are sent to the electronic control unit (position 3) via an electrical conductor (position 5). After receiving the signal from the sensor of the sowing sections, the electronic control unit processes and generates the output signals which control the operation of the electric injectors (opening/closing) (position 1). With an electric pump (position 10), which is located in the tank (position 9), the liquid starter fertilizer is delivered under pressure to the electric injectors.
In the moment of time when it is necessary to apply liquid starter fertilizer, the electronic control unit sends a signal that opens the valve of the electric injector and in that way enables the application of fertilizer (Figure 4). By adjusting the operating parameters of the electronic control unit, it is possible to automatically apply the liquid starter fertilizer in different amounts (L ha −1 ) and by different application techniques.

The Working Principle of the Prototype of the Electronic Device EUKU-01
Schematic and basic construction components of the prototype of the electronic device EUKU-01 (Ciga Factory, Aranđelovac, Republic of Serbia) is shown in Figure 3.  In the moment of time when it is necessary to apply liquid starter fertilizer, the electronic control unit sends a signal that opens the valve of the electric injector and in that way enables the application of fertilizer ( Figure 4). By adjusting the operating parameters of the electronic control unit, it is possible to automatically apply the liquid starter fertilizer in different amounts (L ha −1 ) and by different application techniques. An electric injector with sprayers from the manufacturer 'Lechler' shown in Table 3 was used for the application of the starter fertilizer. By applying different application techniques, the amount of starter fertilizer and the amount of mineral fertilizers on plots from E1 to E15, a total of 5 different treatments were obtained, as shown in Table 4.  An electric injector with sprayers from the manufacturer 'Lechler' shown in Table 3 was used for the application of the starter fertilizer. By applying different application techniques, the amount of starter fertilizer and the amount of mineral fertilizers on plots from E1 to E15, a total of 5 different treatments were obtained, as shown in Table 4.

Data Analysis
A set of 90 observations of corn yield, collected and measured during this research was subjected to statistical data analysis. Mean values and standard deviations were calculated and visualized to obtain an insight into the overall trend. Differences between mean values of achieved yields obtained by applying different treatments were analyzed by two-factor analysis of variance and post-hoc Tukey honest significant difference (HSD) test, where the influence of mineral fertilizer treatment and the influence of liquid starter fertilizer treatment were observed as factors.
All calculations, tests and data visualization were carried out using R programming language, v4.0.0, with a significance level of p = 0.05.

Results
Data on the amount of precipitation and mean monthly air temperatures for the observed sites are shown in Table 5. The conditionally optimal values of precipitation and air temperature [35,36] are presented along with the monthly values for the period 2016-2018. Based on the presented values, it can be stated that the amount of precipitation during July and August at both sites was less than conditionally optimal. Lack of water and average temperatures Agriculture 2020, 10, 347 8 of 13 higher than conditionally optimal led to a decrease in yield during the research in relation to the genetic potential of the cultivated corn hybrid ZP 427.
Corn yield measured in this research ranged from 1.86 t ha −1 up to 6.50 t ha −1 , having overall mean of 3.73 t ha −1 . Mean values of yield observed in different years depending on the liquid fertilizer treatment are presented in Table 6. p-values from two factor analysis of variance for treatments TT 0 -TT 4 and sub treatments T 0 -T 1 are given in Table 7. Statistically significant values are marked with (***) for significance level of 0.001 and (**) for significance level of 0.01. There was no interaction between treatment and sub treatment. Table 7. p-values from two factor analysis of variance for treatments TT 0 -TT 4 and sub treatments T 0 -T 1.

Discussion
Distribution of the yield values is presented with boxplots in Figure 5 (yield for each year), Figure 6 (yield for each treatment with liquid fertilizer) and Figure 7 (yield depending on the treatment with liquid fertilizer by year), showing interquartile range (IQR), minimal, median and maximal values for each data subset presented.
Distribution of the values of the yields obtained by years ( Figure 5) revealed lower values in 2017 for all variants of mineral and starter fertilizer application. The low values of the yields obtained were caused by the insufficient amount of precipitation, significantly less than the conditionally optimal ones, which was especially noticeable during July and August 2017 at both sites. Figure 6 shows the obtained yield values depending on the amount of applied starter fertilizer. There was an increase in the value of yield with an increase in the amount of applied fertilizer TT 1 -TT 2 -TT 3 -TT 4 , regardless of the technique of application ('point' or 'belt') of this fertilizer. It was noticed that 'point' form application and smaller doses of starter fertilizer achieved approximate values of corn yield in relation to the 'belt' form application at higher doses. The difference in the amount of applied starter fertilizer between TT 1 and TT 2 , as well as between TT 3 and TT 4 , was without a statistically significant difference in the values of yields of the observed treatments.
Obtained values of yield depending on the liquid starter fertilizer treatment for the period 2016-2018 are shown in Figure 7. Grain yield per hectare varied statistically significantly depending on the applied fertilization system at both sites during the years of study. The lowest yield values were obtained in treatments where minimal amounts of mineral and starter fertilizers were applied. The sub treatment with mineral fertilizer (T 0 , T 1 , T 2 ) gave statistically significant differences in yields in all three years (with values of p < 0.001). The highest values of achieved yields at both sites were obtained in the sub treatments with the largest amounts of applied mineral fertilizers T 2 ( Table 4). As for the mineral fertilization treatment, subsequent tests (Tukey HSD) showed the presence of a difference in all three sub treatments with p values less than 0.001. The treatment with starter fertilizer showed a statistically significant effect only in 2016 (p = 0.00575). Regarding the 2016 liquid fertilizer treatment, there was a statistically significant difference between TT 3 treatment and TT 0 control (p = 0.01), and TT 4 and TT 0 control (p = 0.005). Table 7. p-values from two factor analysis of variance for treatments TT0-TT4 and sub treatments T0-T1.

Discussion
Distribution of the yield values is presented with boxplots in Figure 5 (yield for each year), Figure 6 (yield for each treatment with liquid fertilizer) and Figure 7 (yield depending on the treatment with liquid fertilizer by year), showing interquartile range (IQR), minimal, median and maximal values for each data subset presented. Distribution of the values of the yields obtained by years ( Figure 5) revealed lower values in 2017 for all variants of mineral and starter fertilizer application. The low values of the yields obtained were caused by the insufficient amount of precipitation, significantly less than the conditionally optimal ones, which was especially noticeable during July and August 2017 at both sites. Figure 6 shows the obtained yield values depending on the amount of applied starter fertilizer. There was an increase in the value of yield with an increase in the amount of applied fertilizer TT1-TT2-TT3-TT4, regardless of the technique of application ('point' or 'belt') of this fertilizer. It was noticed that 'point' form application and smaller doses of starter fertilizer achieved approximate values of corn yield in relation to the 'belt' form application at higher doses. The difference in the amount of applied starter fertilizer between TT1 and TT2, as well as between TT3 and TT4, was without a statistically significant difference in the values of yields of the observed treatments. Obtained values of yield depending on the liquid starter fertilizer treatment for the period 2016-2018 are shown in Figure 7. Grain yield per hectare varied statistically significantly depending on the applied fertilization system at both sites during the years of study. The lowest yield values were obtained in treatments where minimal amounts of mineral and starter fertilizers were applied. The sub treatment with mineral fertilizer (T0, T1, T2) gave statistically significant differences in yields in all three years (with values of p < 0.001). The highest values of achieved yields at both sites were obtained in the sub treatments with the largest amounts of applied mineral fertilizers T2 ( Table 4). As for the mineral fertilization treatment, subsequent tests (Tukey HSD) showed the presence of a difference in all three sub treatments with p values less than 0.001. The treatment with starter fertilizer showed a statistically significant effect only in 2016 (p = 0.00575). Regarding the 2016 liquid fertilizer treatment, there was a statistically significant difference between TT3 treatment and TT0 As expected, the highest overall mean values of yield were obtained in experiments where starter fertilizer was applied along with mineral fertilizer (NPK). These values were recorded in each year of the research on both experimental sites. The increase in corn yield (mean values), in comparison with control plot, was 7.9-17.1%, depending on the applied liquid starter fertilizer treatments (TT 1 , TT 2 , TT 3 or TT 4 ). Similar results were reported by [39][40][41][45][46][47]. It was noted that, during the all three years of research period, there is no significant differences between overall mean values of yield in treatments TT 1 and TT 2 , i.e., TT 3 and TT 4 . The highest difference in yield of 2.5% appeared in 2018 between the treatments TT 1 and TT 2 . Respectively, the highest difference in yield of 4.1% appeared in 2018 between the treatments TT 3 and TT 4 (see Table 2). Experimental results showed that with the application of 30% less of starter fertilizer in the 'point' form, the achieved values of yield were approximately the same as the values as in the case of applying a larger amount in the 'belt' form.
Obtained values of yield depending on the liquid starter fertilizer treatment for the period 2016-2018 are shown in Figure 7. Grain yield per hectare varied statistically significantly depending on the applied fertilization system at both sites during the years of study. The lowest yield values were obtained in treatments where minimal amounts of mineral and starter fertilizers were applied. The sub treatment with mineral fertilizer (T0, T1, T2) gave statistically significant differences in yields in all three years (with values of p < 0.001). The highest values of achieved yields at both sites were obtained in the sub treatments with the largest amounts of applied mineral fertilizers T2 (Table 4). As for the mineral fertilization treatment, subsequent tests (Tukey HSD) showed the presence of a difference in all three sub treatments with p values less than 0.001. The treatment with starter fertilizer showed a statistically significant effect only in 2016 (p = 0.00575). Regarding the 2016 liquid fertilizer treatment, there was a statistically significant difference between TT3 treatment and TT0 control (p = 0.01), and TT4 and TT0 control (p = 0.005).

Conclusions
The impact of application technique and rate of liquid starter fertilizer applied with a novel device on the production of corn was presented in this study. It shows that the yield results were obtained by applying mineral fertilizers (T 0 , T 1 and T 2 ) and different application techniques and amounts of starter fertilizer (TT 0 , TT 1 , TT 2 , TT 3 and TT 4 ). Starter fertilizer was applied in the root system range of freshly germinated plants in the 'belt' and 'point' forms at different quantities (35,50,70, and 100 L ha −1 ), according to the two-by-two placement method, which led to intensive plant growth in the initial stages of development. The field trial was set up at two sites (two different land types), in the conditions of the natural water regime of the soil during the three vegetation seasons in the period 2016-2018. For this purpose, a prototype of the electronic device EUKU-01 was designed to provide a control of the seeding and fertilizing process. Data were statistically analyzed by two-factor analysis of variance, where the influence of mineral fertilizer treatment and the influence of liquid starter fertilizer treatment were observed as factors.
Corn yield measured in this research ranged from 1.86 t ha −1 up to 6.50 t ha −1 , having an overall mean of 3.73 t ha −1 . The lower values of the yields obtained were in 2017 and were caused by the insufficient amount of precipitation, significantly less than the conditionally optimal ones. The increase in the amount of applied mineral fertilizers, increased the value of yield. Grain yield per hectare varied statistically significantly depending on the applied fertilization system at both sites during all the years of study. The highest overall mean values of yield were obtained in experiments where starter fertilizer was applied along with mineral fertilizer. The increase in corn yield (mean values), in comparison with control plot, was 7.9-17.1%, depending on the applied liquid starter fertilizer treatments. The highest difference in yield of 2.5% appeared in 2018 between the treatments TT 1 and TT 2 , i.e., 4.1% in 2018 between the treatments TT 3 and TT 4 . Results obtained showed that with the application of 30% less amount of starter fertilizer in the 'point' form, achieved values of yield were approximately the same as the values in the case of applying a larger amount in the 'belt' form. The application of liquid