*Article* **Impact of Maize–Mushroom Intercropping on the Soil Bacterial Community Composition in Northeast China**

#### **Xiaoqin Yang 1, Yang Wang 1,\*, Luying Sun 1, Xiaoning Qi 1, Fengbin Song <sup>1</sup> and Xiancan Zhu 1,2,\***


Received: 18 August 2020; Accepted: 30 September 2020; Published: 7 October 2020

**Abstract:** Conservative agricultural practices have been adopted to improve soil quality and maintain crop productivity. An efficient intercropping of maize with mushroom has been developed in Northeast China. The objective of this study was to evaluate and compare the effects of planting patterns on the diversity and structure of the soil bacterial communities at a 0–20 cm depth in the black soil zone of Northeast China. The experiment consisted of monoculture of maize and mushroom, and intercropping in a split-plot arrangement. The characteristics of soil microbial communities were performed by 16S rRNA gene amplicom sequencing. The results showed that intercropping increased soil bacterial richness and diversity compared with maize monoculture. The relative abundances of Acidobacteria, Chloroflexi, Saccharibacteria and Planctomycetes were significantly higher, whereas Proteobacteria and Firmicutes were lower in intercropping than maize monoculture. Redundancy analysis suggested that pH, NO3 <sup>−</sup>-N and NH4 <sup>+</sup>-N contents had a notable effect on the structure of the bacterial communities. Moreover, intercropping significantly increased the relative abundance of carbohydrate metabolism pathway functional groups. Overall, these findings demonstrated that intercropping of maize with mushroom strongly impacts the physical and chemical properties of soil as well as the diversity and structure of the soil bacterial communities, suggesting this is a sustainable agricultural management practice in Northeast China.

**Keywords:** 16S rRNA; planting pattern; soil chemical properties; soil microbial community

#### **1. Introduction**

Maize (*Zea mays* L.) is not only one of the three major food crops in the world, but also a high-quality animal feed and an important industrial raw material. Maize production in China accounts for 21.9% of total maize output in the world. The northeast area is the largest maize producer in China, particularly in the Jilin province, which accounts for 1/8th of maize production in China [1]. However, the maize planting pattern in Northeast China was mainly based upon conventional agricultural practices (e.g., monoculture cropping, tillage and removal of crop residues) that have caused soil erosion, fertility decline, and loss of soil biodiversity [2]. Therefore, conservation and sustainable agricultural practices, such as intercropping, no tillage and mulching, have been adopted to improve soil quality and increase crop productivity.

Intercropping, a method of simultaneously planting two or more crops on the same field, has been practiced in many countries for several decades [3]. It increases the productivity per unit of land through better utilization of resources, control of disease and reducing soil erosion [4]. Soil microbial activity, nutrient cycling, decomposition of organic matter, and physical and chemical properties can be changed in intercropping systems [5,6]. Soil organic matter, reflecting the physical, chemical and biological properties of soil, is an effective indicator of soil quality [2,4], which is defined as the capacity of a soil to function within ecosystem and land-use boundaries to sustain biological productivity, maintain environmental quality, and promote plant and animal health [7]. Mushroom compost is used as a soil conditioner for the growth of soybean, and the substrate pH for growing soybean is increased from acidity to neutrality [8]. The residual fungi and a large number of mycelium are incorporated into the soil, making the soil loose, which improves the physical and chemical properties of the soil, and increases the content of various nutrients in the soil in the maize–mushroom intercropping system [9]. Soil microorganisms play crucial roles in soil ecosystem functioning, as they are involved in soil nutrient cycling and energy flows, micro-ecology regulation and soil sustainable productivity [4,10]. In the soil ecosystem, soil bacteria are crucial decomposers that metabolize organic matter by secreting specific extracellular enzymes to break down large organic molecules into monomers, which are then available for plant uptake [11,12]. The cultivation of continuous monoculture often results in the accumulation of soil-borne pathogens, the growth inhibition of beneficial bacteria [13], and a decrease of bacterial diversity, ultimately [14]. Pear/mushroom intercropping significantly increased the number and biomass of cultivable microorganisms in the 0–40-cm soil layer and had a marked effect on the soil fertility of the pear garden and the quality of the pear fruit [15].

Oyster mushroom (class Basidiomycetes) is a fungus that is an excellent source of human nutrients (e.g., vitamins, minerals and micro nutrients) and has medicinal values [16]. It can be used as an intercropping crop to promote the efficient uptake of nutrients and resources [8]. For example, the mushroom (*Plearotus* spp.) intercropped with field-grown faba bean (*Vicia faba* L.) increased the faba bean dry seed yield and mushroom basidiocarp yield compared to sole cultivation [16]. Maize–mushroom intercropping systems have been previously studied and were proved to improve maize yield and land equivalent ratio [9,17]. However, few studies have attempted to investigate the effects of maize–mushroom intercropping on soil physical and chemical properties and the diversity and composition of soil microbial community. Therefore, the objective of this study was to explore the effect of maize–mushroom intercropping on the soil bacterial community in Northeast China. Recently, we developed an integrated agricultural management practice involving wide–narrow-row spacing alternation (160 + 40 cm), adoption of no tillage, and mulching by crop residues in maize cropping system in the Northeast China plain, resulting in an improvement of soil quality, reduced soil erosion, and a more sustainable use of cultivated land [18]. Furthermore, in order to use land resources with high efficiency, control disease, and increase the income of farmers, maize–mushroom intercropping was developed based on the integrated agricultural practice. We hypothesized that the intercropping could lead to (i) a marked effect on the structure and diversity of the soil bacterial community and (ii) an improvement of soil quality compared with monocultures.

#### **2. Materials and Methods**

#### *2.1. Site Description and Experimental Design*

The integrated agricultural management practice was initiated in spring 2012 and the maize–mushroom intercropping under integrated agricultural practice experiment was initiated in spring 2014 at Changchun Agricultural Experimental Field of the Northeast Institute of Geography and Agroecology, Jilin Province, China (43◦59 54" N, 125◦23 57" E). The climate is a north temperate continental monsoon. The mean annual air temperature is 7.2 ◦C. The mean annual precipitation is 530.5 mm. The frost-free period is 138 days. Snow cover usually occurs from November to April. The soil is a typical silty clay loam (classified as Mollisols), which is highly fertile, inherently productive and suitable soil for cultivation in the world [19], with organic matter 3.1%, total nitrogen 1.49 g kg<sup>−</sup>1, total phosphorus 0.59 g kg<sup>−</sup>1, total potassium 21.53 g kg−1, available nitrogen 204.54 mg kg−1, available phosphorus 9.43 mg kg−1, effective potassium 125.84 mg kg−1, soil bulk density 1.19 g cm−3, cation exchange capacity 23.3 cmol kg<sup>−</sup>1, and pH 6.71.

The experiment was conducted in a plot comparison experiment with three treatments including (1) maize–mushroom intercropping, (2) maize monocultures, and (3) mushroom monocultures. Each plot was 20 m long and 8 m wide with four plot replicates for each treatment. The cultivars were maize Liangyu 99 and oyster mushroom (provided by Liaoning Sanyou Agricultural Biotechnology Company). The maize seeds were sown in late April and harvested in late September, and mushroom sticks were planted in early July and harvested in late September every year. In the maize monoculture treatment, the wide–narrow row spacing was 160 and 40 cm, respectively; no tillage was implemented; the maize stalks were cut at 35 cm above the ground, and then maize stalks were left in the field. Maize was planted on a narrow line of 40-cm spacing, with a plant spacing of 15 cm and a planting density of 6.5 <sup>×</sup> <sup>10</sup><sup>4</sup> plants ha<sup>−</sup>1. The intercropping of maize with mushroom was based on the integrated agricultural management practice. In the intercropping and maize monoculture, maize seeds were sown in 40-cm, narrow rows using a small jukebox in late April. During the maize spinning period, a 10-cm-deep and 40-cm-wide sulcus was dug in advance to place mushroom sticks in the intercropping. Then, mushroom sticks were evenly placed in the sulcus, covered in 2–3 cm soil (Figure 1). The distance between mushroom and maize, mushroom and high stubble both were 27 cm. The mushroom row spacing of mushroom monoculture treatment was 160 cm. A based controlled release compound fertilizer (resin coating) of 600 kg ha−<sup>1</sup> (total nutrient <sup>≥</sup> 53%, Zn <sup>≥</sup> 2%) was applied to all treatments (as the base of the fertilizer) before rotating the soil (using the small rotating machine). The type of controlled release nutrient was nitrogen, and the amount of controlled release nutrient was greater or equal to 8%. In mid-May, the compound fertilizer was added once. Throughout the growth period of mushroom, intermittent spray irrigation was used to maintain the optimum moisture for mushroom production.

**Figure 1.** Schematic illustration of row placement of (**a**) maize–mushroom intercropping, (**b**) maize monoculture and (**c**) mushroom monoculture. Red arrows indicate the sampling point. The unit of distances is the centimeter (cm).

#### *2.2. Soil Sampling*

On September 20 2017 (After the mushrooms had been planted for 70 days), soil core samples (5.5 cm in diameter, 20 cm in depth, repeated four times) from each treatment were taken randomly from the maize rows (10 cm from the maize rows) of maize monoculture and intercropping treatment, and high stubble rows (10 cm from high stubble rows) of mushroom treatment. In total, there were 12 soil samples used for the analysis. All samples were sieved by a 2-mm sieve to remove rocks and were

then separated into two parts: one part air dried to determine soil properties, and another stored at −80 ◦C for DNA extraction.

#### *2.3. Soil Chemical Assays*

Total nitrogen (TN), nitrate nitrogen (NO3 <sup>−</sup>-N), ammonium nitrogen (NH4 <sup>+</sup>-N), were determined by a continuous flow analyzer (San++, Skalar, Breda, Holland). Soil organic matter (SOM) was measured by the method of potassium dichromate heating oxidation-volumetric based on the standard LY/T1237-1999 [20]. Briefly, 0.2 g of air-dried soil was added into 5 mL K2Cr2O7 (0.8 M) and 5 mL concentrated H2SO4 solution and heated on a hot plate (300 ◦C) for 5 min. Three to four drops of phenanthroline indicator were then added, and titrated with FeSO4 (0.4 M).

TN, available nitrogen (AN), NH4 <sup>+</sup>-N, and NO3 −-N were measured by standard methods based on LY/T1228-2015. For TN, 1 g soil was added to 1.8 g catalyst (selenium powder: copper sulfate: potassium sulfate = 1:10:100), mixed with 4 mL concentrated sulfuric acid, and removed on the electric heating plate until the soil became grayish white. The resulting solution was green and was transferred to a 100-mL volumetric flask after cooling. The solution was then diluted to volume with distilled water and shaken for testing. For AN, 2 g soil was mixed with 3 mL of 2% H3BO3, 10 mL of 1 M NaOH solution, and then incubated at 40 ◦C for 24 h. After cooling, the sample was titrated with HCl (0.012 M) until the color changed from blue-green to purple. For NH4 <sup>+</sup>-N and NO3 −-N, 5 g soil was mixed with 25 mL of KCl (2 M), shaken for 30 min, and filtered for testing.

To measure soil pH, 10 g soil was placed in a 100-mL beaker. Then, 50 mL of distilled water (water:soil = 5:1) was added, shaken for 30 min, and measured with a calibrated pH meter (Mettler-Toledo FE 20, Zurich, Switzerland).

#### *2.4. DNA Extraction and MiSeq Sequencing*

DNA was extracted from 0.25 g fresh soil using E.Z.N.A Mag-Bind Soil DNA Kits (OMEGA, Irving, TX, USA) according to the manufacturer's instructions. The DNA extraction quality was measured by 0.8% agarose gel electrophoresis, and the DNA was quantified by an ultraviolet spectrophotometer. The PCR amplification used Q5 high-fidelity DNA polymerase (NEB, Ipswich, MA, USA), and the V3-V4 region of the 16S rDNA genes were amplified using the primers 338F (ACTCCTACGGGAGGCAGCA) and 806R (GGACTACHVGGGTWTCTAAT) [21]. Referring to the preliminary quantitative results of gel electrophoresis, the PCR amplified product was subjected to fluorescence quantification, the fluorescent reagent was Quant-iTPicoGreen dsDNA Assay Kit, and the quantitative instrument was Microplate reader (BioTek, FLx800). A total of 20 pM DNA for each sample were pooled and sequenced in an Illumina MiSeq platform (Illumina, SanDiego, CA, USA) with a 600-cycle kit (2 × 300 bp paired ends). The Miseq sequencing raw data were deposited in the NCBI Sequence Read Archive database, and the project ID is PRJNA432129 and BioSample is SAMN10362720.

#### *2.5. Data Analysis*

The original paired-end sequencing data were exported in a FASTQ format. Sequences were removed if the read length was <150 bp, with a mean quality score <20 [22]. Sequence analysis was performed using QIIME software (Version 1.8.0) [23]. Sequences with ≥97% similarity were assigned to the same operational taxonomic units (OTUs) using UCLUST [24]. In detail, the sequences were merged according to ≥97% similarity into OTUs, and the most abundant sequence in each OTU was chose as the representative OTU sequence. Then, according to the number of sequences of each OTU in each sample, a matrix file of OTU abundance in each sample was constructed (i.e., OTU table). The relative abundance of OTUs with <0.001% of the total reads of all samples were removed [25]. The Greengenes database (Release 13.8) was used to annotate taxonomic information [26].

Data were compared using analysis of variance (ANOVA) in IBM SPSS 23.0 software (SPSS Inc., USA). Alpha diversity (Chao1 and Shannon index) was calculated with QIIME (Version 1.8.0). Pearson's correlation coefficients between soil properties, OTU richness (Chao1 index) and diversity (Shannon index), and bacterial phyla were computed.

Differences of soil bacterial communities based on OTUs between treatments were analyzed using LEfSe [27]. Cluster analysis of soil bacterial communities based on the non-metric multidimensional scaling (NMDS) dissimilarity matrix was performed using QIIME. Permutational multivariate analysis of variance (PERMANOVA; function 'adonis') was adopted to compare community composition on three treatments with QIIME. Redundancy analysis (RDA) was carried out to explore the relationship between soil properties and microbial community composition using the 'vegan' package in the R program (Version 3.3.1). Function predictions were classified into KEGG pathways using PICRUSt (Version 1.1.4) method [28]. In detail, firstly, OTU table was standardized by copy number; the full-length 16S rRNA gene sequence of the tested microbial genome was used to infer the gene function spectrum of their common ancestor; then, the gene function profiles of other untested species in the Greengenes 16S rRNA gene full-length sequence database was inferred and constructed the gene function prediction profiles of the entire lineage of archaea and bacteria; thirdly, the 16S rRNA gene sequence data obtained by sequencing with the Greengenes database was compared to find the "nearest neighbor of the reference sequence" of each sequenced sequence and classified it as a reference OTU; the obtained OTU abundance matrix according to the copy number of the rRNA gene of the nearest neighbor of the reference sequence was corrected; finally, the microbial composition data to the known gene function profile database was "mapped" to realize the prediction of the metabolic function of the microbial communities. The KEGG pathways statistical analysis was implemented using SPSS.

#### **3. Results**

#### *3.1. Soil Chemical Properties*

The planting pattern was found to significantly affect soil pH and the contents of AN, NO3 −-N, NH4 <sup>+</sup>-N, and SOM (Table 1). The soil pH and SOM were markedly increased in intercropping, whereas the contents of AN, NO3 <sup>−</sup>-N and NH4 <sup>+</sup>-N were decreased, compared with maize monoculture. Soil pH was higher, whereas AN, NO3 <sup>−</sup>-N and NH4 <sup>+</sup>-N contents were lower in mushroom monoculture than maize monoculture.

**Table 1.** Soil chemical properties under maize monoculture, mushroom monoculture and maize–mushroom intercropping.


Values are means ± standard errors. Mean values in each column followed by the different letters are significantly different (*p* < 0.05) according to Tukey's test. The number of replicates per treatment is 4 (*n* = 4). TN, AN, NO3 −-N, NH4 <sup>+</sup>-N and SOM represent total nitrogen, available nitrogen, nitrate nitrogen, ammonium nitrogen and soil organic matter, respectively.

#### *3.2. Soil Bacterial Community Diversity*

Sequencing of the 16S rRNA gene fragment from 12 soil samples produced a total of 462,813 sequences (Table S1). After filtration, alignment, pre-clustering and removal of chimeric sequences and singletons, 359,341 sequences were obtained. They were assigned into 18,519 OTUs.

Intercropping and mushroom monoculture had higher bacterial OTU richness and Shannon index compared with maize monoculture (Figure 2). There was a striking positive relationship between Shannon index and pH and SOM, whereas there was a negative association between Shannon index and the contents of AN, NO3 <sup>−</sup>-N, NH4 <sup>+</sup>-N (Table S2).

**Figure 2.** Operational taxonomic units (OTUs) richness and Shannon index of soil bacterial community under maize monoculture, mushroom monoculture and maize–mushroom intercropping. The error bars represent standard errors (SE). Different letters indicate significant different (*p* < 0.05) according to Tukey's test. The number of replicates per treatment is 4 (*n* = 4).

#### *3.3. Soil Bacterial Community Structure*

In order to characterize the effect of planting pattern on soil bacterial communities, the relative abundances at phylum, class, order, family and genus levels were analyzed. Then, 16S rRNA gene amplicon sequencing showed that the dominant bacterial phyla were Proteobacteria, Chloroflexi, Acidobacteria, Actinobacteria, Firmicutes, Gemmatimonadetes, Cyanobacteria, Saccharibacteria, Bacteroidetes, Nitrospirae, Planctomycetes, Verrucomicrobia and Parcubacteria (Figure 3), and these groups accounted for over 88–95% of the sequences. Moreover, soil chemical properties were closely correlated with the relative abundance of some dominant microbial phyla groups (Table S2). There was a marked positive association between soil pH and Chloroflexi, Acidobacteria, Planctomycetes, Verrucomicrobia, and Parcubacteria, whereas the relationships were negative between the contents of AN, NO3 <sup>−</sup>-N, NH4 <sup>+</sup>-N and those dominant microbial phyla.

In total, 13 phyla, 31 classes, 64 orders, and 84 families in the bacterial community were significantly affected by planting pattern. At the phylum level, planting pattern significantly altered the relative abundance of Chloroflexi, Acidobacteria, Saccharibacteria, Planctomycetes, Armatimonadetes, Fusobacteria, Euryarchaeota, WS6, Elusimicrobia, Peregrinibacteria, WWE3 and Gracilibacteria. Intercropping had higher relative abundance of Acidobacteria, Chloroflexi, and Planctomycetes, and lower relative abundance of Proteobacteria and Firmicutes compared with maize monoculture treatment (Figure 3). Intercropping had higher relative abundance of Saccharibacteria, and lower Cyanobacteria than mushroom monoculture. The distribution of related genera varied between maize monoculture, mushroom monoculture and intercropping soils. *Pseudomonas*, *Sphingomonas*, *Lactobacillus* and *Rhodanobacter* were the most abundant genera across all soil samples, representing 5.57%, 2.83%, 0.19%, and 3.74% of all classified sequences in intercropping, 17.24%, 9.28%, 10.79%, and 2.05% in maize monoculture and 6.64%, 1.59%, 0.14%, and 1.05% in mushroom monoculture, respectively. *Gemmatimonas*, *Cystobacteraceae*, *Marmoricola*, *Bradyrhizobium*, *Streptomyces*, *Rhodoglobus*, *Nakamurella*, *Frateuria*, *Sphingobium*, *Woodsholea*, and *Oribacterium* showed an increased relative abundance in

intercropping, while the relative abundance of *Streptomycetaceae* decreased in intercropping compared with maize monoculture (Figure 4).

**Figure 3.** Relative abundances of the dominant phyla of bacteria at 0-20-cm soil depths under maize monoculture, mushroom monoculture and maize–mushroom intercropping. The number of replicates per treatment is 4 (*n* = 4).

**Figure 4.** Cladogram of bacteria at 0–20-cm soil depths under maize monoculture, mushroom monoculture and maize–mushroom intercropping. The Cladogram shows the hierarchical relationship of all classification units from the phylum to the genus (from the inner circle to the outer circle). The node size corresponds to the average relative abundance of the classification units. The blue, green and red node represents that the difference of bacterial relative abundance between groups is significant. The letters identify the taxon name that has a significant difference between groups. The number of replicates per treatment is 4 (*n* = 4).

#### *3.4. Comparative Analysis of Soil Bacterial Community*

Based on the NMDS dissimilarity matrix, hierarchical cluster analysis for investigated the beta diversity of bacterial communities showed that soil bacterial communities were affected by planting patterns (Figure 5). PERMANOVA analysis indicated that soil bacterial community composition was significantly affected by maize monoculture, mushroom monoculture and intercropping treatments. The RDA was performed to determine the strength of the association between the soil bacterial community and soil physical, chemical properties. RDA revealed a strong difference between maize monoculture, mushroom monoculture and intercropping soils (Figure 6). The first two canonical axes

are responsible for 45.73% of variance (25.42% by RDA1 axis and 20.31% by RDA2 axis). The RDA indicated that pH, NO3 <sup>−</sup>-N and NH4 <sup>+</sup>-N contents had an extremely significant influence on the structure of the bacterial community (*p* < 0.05) (Table S3).

**Figure 5.** Hierarchical cluster analysis of soil bacterial communities based on the NMDS dissimilarity matrix among maize monoculture, mushroom monoculture and intercropping. Permutational multivariate analysis of variance (PERMANOVA) was adopted to compare community composition on three treatments. The number of replicates per treatment is 4 (*n* = 4).

**Figure 6.** Redundancy analysis (RDA) of soil bacterial community structure and soil properties under maize monoculture, mushroom monoculture and intercropping. Soil factors indicated in blue text include pH, contents of organic matter (SOM), total nitrogen (TN), available nitrogen (AN), nitrate nitrogen (NO3 <sup>−</sup>-N) and ammonium nitrogen (NH4 <sup>+</sup>-N). The circles are the RDA scores of the samples and the arrows are the scores of the soil variables by RDA. The number of replicates per treatment is 4 (*n* = 4).

#### *3.5. Metabolism of Soil Microbial Community*

Potential metabolism were assigned to predicted functional annotation of protein sequences. The KEGG metabolic pathway difference analysis of soil bacteria revealed significant changes in 11 metabolic networks between the groups of planting patterns. The analysis showed that intercropping significantly increased the relative abundance of carbohydrate metabolism pathway functional groups compared with maize monoculture (Figure 7). In addition, the relative abundance of glycan biosynthesis and other secondary metabolites functional groups in intercropping and mushroom monoculture was higher compared with maize monoculture (*p* < 0.05) (Figure 7 and Table S4).

**Figure 7.** The KEGG metabolic pathway difference analysis of bacteria at 0–20-cm soil depths under maize monoculture, mushroom monoculture and intercropping. The number of replicates per treatment is 4 (*n* = 4).

#### **4. Discussion**

#### *4.1. Soil Physicochemical Properties*

Nutrients such as carbon, nitrogen, phosphorus and potassium, are essential for proper plant growth. Soil carbon and nitrogen contents are the most sensitive indicators of soil quality [29], and it has been suggested that below ground interspecific interactions improved soil nitrogen supply capacity [30], mobilization [31] and increased TN in intercropping [32]. The results of the current study have shown that contents of AN, NO3 <sup>−</sup>-N, NH4 <sup>+</sup>-N in intercropping treatment were no higher than maize monoculture treatment. Below ground interactions through intercropping could affect N-cycling [30,33]. Moreover, soil nitrate nitrogen is dynamic, and is influenced by soil particle distribution, soil depth, and precipitation, and varies during crop growth and development [34]. Soil nitrate would likely decline faster in intercropping and mushroom monoculture treatments in comparison with maize monoculture, mostly due to decomposition of mineral nutrients by saprotrophs and leaching losses [35], using intermittent spray irrigation to maintain the optimum moisture for

mushroom production. Vieira et al. [36] also showed that nitrogen losses after 10 and 15 days impacted by mushroom yield. In this study, intercropping treatment significantly increased the contents of SOM compared with maize monoculture. This is consistent with previous studies that demonstrated that soil organic carbon fraction, carbon pool management index and soil carbon sequestration were improved under intercropping [4,37,38].

#### *4.2. Bacterial Community Diversity*

The soil microbial community, a biomarker indicator of soil quality and ecosystem processes [39], is very sensitive to vegetation. On the other hand, it can also strongly affect plant growth and yield formation [40]. Our study revealed that intercropping significantly increased the OTU richness and diversity of the bacterial community compared to maize monoculture treatment. Consistent with our findings, Fu et al. [4] reported that maize–soybean intercropping had higher Shannon index compared with monocultures. Qin et al. [41] demonstrated that maize–potato intercropping increased the carbon source utilization rate and diversity of the microbial community. In addition, the bacterial Shannon index was correlated with contents of N, SOM and pH, suggesting that the improvements in C and N source utilization were beneficial to increased soil bacterial diversity, such as species richness [12]. Moreover, soil bacterial community could be changed by pH with a higher bacterial diversity in neutral soil than acidic soil, which a significant positive association between Shannon index and pH was found in this study.

#### *4.3. Bacterial Community Structure*

Microbial community composition has large effects on organic matter dynamics and nutrient cycling, and can influence soil function and ecosystem sustainability [42]. In total, 46 phyla, 168 classes, 356 orders, 605 families and 1207 genera of bacterial communities were obtained in our samples. Within the thirteen dominant bacterial phyla, Proteobacteria was the most abundant bacterial phylum, which was consistent with the results of mulberry and alfalfa intercropping system [43]. The major microbial phyla, such as Chloroflexi, Acidobacteria, Actinobacteria, identified in this study are often observed in other soils, though the relative abundance was different [44,45].

Previous studies have indicated that the dominant bacterial phyla could be changed by manipulating planting patterns and plant species [4,12,43]. In the present study, the bacterial community structures among intercropping and maize and mushroom monoculture treatments were significantly different. Planting patterns significantly affected the four dominant and eight other bacterial phyla. The relative abundance of Chloroflexi, Acidobacteria, Planctomycetes and Saccharibacteria were higher in intercropping than maize or mushroom monoculture treatments. Anaerolineae, a dominant class of Chloroflexi, has been thought to be ubiquitous and to play important roles in ecosystems [46]. The greater abundance of Chloroflexi in the intercropping than maize monoculture treatment likely indicated that intercropping would better coordinate soil ecosystems [47]. Acidobacteria is an acidophilic and oligotrophic chemoorganotrophic bacterium; Planctomycetes play possible role in the evolution of the methane cycle [48]. Saccharibacteria play a role in the degradation of various organic compounds as well as sugar compounds under aerobic, nitrate reducing, and anaerobic conditions [49]. Although the phylum Proteobacteria was not affected by planting pattern, Gammaproteobacteria, the most dominant class of Proteobacteria in our study, had a prominent higher abundance under intercropping than maize monoculture treatment, stimulated by higher SOM and lower nitrate nitrogen contents of intercropping treatment [50]. *Pseudomonas* is one of the widely distributed plant growth promoting rhizobacteria [51]. The result revealed that the soil *Pseudomonas* community was affected by planting pattern. Consistent with previous studies, the altered plant species affected the microbial community composition [52].

In this study, the relative abundance of Actinobacteria in intercropping and maize monoculture treatment were significant different at class level. At genus level, *Gemmatimonas*, *Streptomyces*, *Nakamurella* and *Frateuria* had greater abundance in intercropping soils. *Actinobacteria* have a critical

role in decomposition of soil organic materials, such as cellulose, chitin and polysaccharides [53]. Streptomyces could produce bioactive secondary metabolites which show antifungals and antivirals biological activities [54], and have more efficient secretion mechanisms which could promote protein solubilization [55]; *Nakamurella* is able to accumulate polysaccharides [56]; *Frateuria* is able to enhance potassium uptake efficiently in plants and has been found to increase biomass and nutrient content [57]. In addition, *Gemmatimonas*, the dominant genus of Gemmatimonadetes, is involved in modulating carbon and nitrogen intake, decomposing polyaromatic carbon and promoting plant development [58]. The greater abundance of *Actinobacteria*, *Streptomyces*, *Nakamurella*, *Frateuria* and *Gemmatimonas* likely be attributed to the effects of secondary metabolites such as carbohydrates, free amino acids, and nucleotides produced by mushroom during metabolism [59]. Our study showed that the relative abundance of glycan and secondary metabolites functional groups in intercropping and mushroom monoculture significantly increased compared with maize monoculture. This is consistent with the other research that intercropping lead to changes in plants accumulation of minerals and secondary metabolites [60].

To investigate the relationships between soil microbial community structure and measured soil variables in the maize monoculture, monoculture mushroom and intercropping systems, we analyzed the dominant bacterial phyla and OTUs data using Pearson's correlation and RDA. The soil variables have a substantial impact on the dominant bacterial phyla. In this study, SOM, TN, AN, NO3 −-N, NH4 <sup>+</sup>-N, and pH had positive/negative correlations with the dominant bacterial phyla. For example, the abundance of Chloroflexi and Planctomycetes were positively correlated with SOM content, which indicated that SOM correlated with the relative abundance of these bacteria [4]. SOM may play non-negligible roles in influencing on the soil microbial community structure by affecting the metabolism of soil microbes [61]. The output of RDA with soil variation in bacterial community indicated that pH plays critical roles in the structure of the bacterial community. This was in accordance with other studies that indicated that soil pH was a major factor in determining the structure of the soil bacterial community [12,61].

#### **5. Conclusions**

In total, 13 phyla, 31 classes, 64 orders, 84 families in the bacterial community were significantly affected by planting pattern. The results revealed that SOM, TN contents, Shannon index and the relative abundance of Chloroflexi, Acidobacteria, Saccharibacteria, Planctomycetes and carbohydrate metabolism pathway functional groups were significantly increased in intercropping compared with maize monoculture treatment. Moreover, soil chemical properties closely correlated with OTU richness, Shannon index and the relative abundance of Chloroflexi, Acidobacteria, Planctomycetes, Verrucomicrobia, Parcubacteria phyla groups. Our study demonstrates that intercropping of maize with mushroom affected the physical and chemical properties of the soil, and altered the structure and diversity of the soil microbial community. These results suggest that this crop production system could be a sustainable efficient agricultural management practice in Northeast China.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4395/10/10/1526/s1, Table S1: Numbers of bacterial sequences and OTUs identified by 16S rRNA gene sequencing, Table S2: Pearson's correlation coefficients between relative abundances of dominant bacterial phyla and soil properties, Table S3: The environmental vectors onto two ordination of redundancy analysis, Table S4: The analysis of variance of the KEGG metabolic pathway.

**Author Contributions:** Conceptualization, F.S. and X.Z.; methodology, X.Y.; software, X.Y. and X.Z.; investigation, X.Y., L.S. and X.Q.; data curation, X.Y.; writing—original draft preparation, X.Y.; writing—review and editing, Y.W. and X.Z.; supervision, X.Q.; project administration, Y.W. and X.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the "One-Three-Five" Strategic Planning Program of Chinese Academy of Sciences, grant number IGA-135-04.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Compost as an Option for Sustainable Crop Production at Low Stocking Rates in Organic Farming**

**Christopher Brock 1,\*, Meike Oltmanns 1, Christoph Matthes 2, Ben Schmehe 2, Harald Schaaf 3, Detlef Burghardt 3, Hartmut Horst <sup>3</sup> and Hartmut Spieß <sup>2</sup>**


**Abstract:** Mixed-crop-livestock farms offer the best conditions for sustainable nutrient management in organic farming. However, if stocking rates are too low, sustainability might be threatened. Therefore, we studied the development of soil organic matter and nutrients as well as crop yields over the first course of a new long-term field experiment with a mimicked cattle stocking rate of 0.6 LU ha−1, which is the actual average stocking rate for organic farms in Germany. In the experiment, we tested the effects of additional compost application to improve organic matter supply to soils, and further, potassium sulfate fertilization for an improved nutrition of fodder legumes. Compost was made from internal resources of the farm (woody material from hedge-cutting). Soil organic matter and nutrient stocks decreased in the control treatment, even though yield levels, and thus nutrient exports, were comparably low. With compost application, soil organic matter and nutrient exports could be compensated for. At the same time, the yields increased but stayed at a moderate level. Potassium sulfate fertilization further improved N yields. We conclude that compost from internal resources is a viable solution to facilitate sustainable organic crop production at low stocking rates. However, we are aware that this option does not solve the basic problem of open nutrient cycles on the farm gate level.

**Keywords:** long term field experiment; sustainable crop production; nutrient balances; legume nutrition

#### **1. Introduction**

Soil organic matter is recognized as a key factor of soil fertility [1]. For this reason, the supply of soils with organic matter was always a major concern in organic agriculture. Meanwhile, it was shown that organic farming in fact leads to higher soil organic matter levels than conventional management [2]. However, a sufficient supply of soils with organic matter is not an effect of organic farming per se, but of the specific structure of organic systems. Leithold et al. [3] emphasized that fodder legumes and cattle manure are the basic factors for a sufficient supply of soils with organic matter. These factors must balance the loss of soil organic matter in turnover. If the supply of soil with organic matter is too low to meet the specific requirements, SOM levels might decrease even under organic management. This situation was observed in the OAFEG long-term field experiment in Germany that is designed to study the effects of mixed, as compared to stockless organic farming [4]. Under the conditions of this experiment, SOM stocks increased under the mixed farming treatment, but stayed unchanged or even decreased under the two stockless treatments. In a modeling study, Brock et al. [5] calculated that the actual average soil organic matter balance of organic farming in Germany was slightly negative, as the mean

Matthes, C.; Schmehe, B.; Schaaf, H.; Burghardt, D.; Horst, H.; Spieß, H. Compost as an Option for Sustainable Crop Production at Low Stocking Rates in Organic Farming. *Agronomy* **2021**, *11*, 1078. https://doi.org/ 10.3390/agronomy11061078

**Citation:** Brock, C.; Oltmanns, M.;

Academic Editors: Nikolaos Monokrousos and Efimia M. Papatheodorou

Received: 14 April 2021 Accepted: 17 May 2021 Published: 27 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

animal stocking rate was only 0.63 LU per ha at that time. Even though this result should not be overrated due to the high uncertainty of the calculation, it seems necessary to further study soil organic matter changes under organic management with low stocking rates or even stockless systems. If manure availability is too low, farmers will need to utilize further sources of organic matter. Here, green manure and compost are the most important options.

The demand for organic matter to maintain or even increase soil organic matter stocks is dependent on site conditions, management history, and actual management [6]. Organic inputs of plant roots and residues, animal manure, and other material must balance the loss of organic matter in turnover. As organic matter supply and turnover are directly linked to N supply in organic farming, the demand for organic matter is greater with higher yield levels (of non-legumes), due to the export of mineralized N [3].

Compost is reported as a viable option to increase soil organic matter and soil health [7,8]. In principle, composting is the biological decomposition of organic residues [9]. Compost can be made from different substrates, e.g., municipal waste, sewage sludges, plant residues/green waste, farmyard manure, or biogas production residues. Farm compost, as applied in the field experiment reported in this study, is carried out individually in farms, depending on the available materials. However, Lehtinen et al. [10] found that impacts on soil properties and crop yields were not significantly different between the composts made from municipal waste, sewage sludge, green waste, and farmyard manure in a long-term field experiment, even though the macronutrient inputs differed. Further, microbial biomass and the composition of the microbial community differed between the treatments [11].

In general, compost application builds up soil organic matter [12] and enhances crop yields moderately in the short run [13]. In the long run, the build-up of soil organic matter further improves the growing conditions for arable crops and thereby further increases yields [14,15].

The biological N fixation (BNF) is an important source of N for organic crop rotations because mineral sources of N fertilizers that are allowed for organic farming are limited. Especially in organic agriculture, BNF is preferred due to different advantages, as compared to mineral N sources like higher N use efficiency of the plants and decreased volatilization, denitrification, and leaching [16]. Therefore, nitrogen fixing legumes like clover and lucerne are usually placed at the beginning of organic crop rotations and act as drivers for the subsequent crops. However, clover and lucerne react particularly sensitively to the deficiencies of P, K, and S. Although several processes and mechanisms about the dependency of legume growth to the listed elements remain unclear [17] it is evident that a good supply improves crop growth and health. It is also known that legumes that acquire N by BNF have a higher demand of P, K, and S, as compared to those that rely on soil N only [18,19]. It is generally accepted that when the host plant growth is reduced due to deficiencies of P, K, or S, an N-feedback is triggered so that the nodule development and activity is reduced. This mechanism can also be induced by plant diseases and pathogens, as well as abiotic stresses like drought, toxic levels of salt, or heavy metals [20–22].

In this study, we showed the development of crop yields and soil nutrients and organic matter over the first crop rotation in a long-term field experiment, under conditions of organic farming (more specifically—biodynamic farming). The experiment mimicked a mixed farm with a stocking rate of 0.6 LU cattle per hectare, which corresponded to the average stocking rate of organic farms in Germany. In this experiment, we compared a fertilization regime that was based on the available cattle manure with a regime that additionally utilized a farm compost made from available plant residues on the farm. Further, we examined the effect of potassium sulfate application, which was owed to the fact that the experiment was located on a potassium-fixing soil.

As the field experiment is still in an early stage, we can only study the short-term effects and development factors, rather than development trends. In this stage, we expect the positive effects of compost application on crop yields, and increased biological N fixation rates in legumes with potassium sulfate application. Further, we want to study the impact of the fertilization regimes on soil nutrient and organic matter balances. This is of high relevance in organic farming, as crop production is largely dependent on soil fertility.

#### **2. Materials and Methods**

We analyzed crop yields and the development of nutrient (N, P, K, S) and organic matter stocks in the soils under the four treatments, in a long-term field experiment on a luvisol, under conditions of biodynamic farming. Further, we calculated nutrient and soil organic matter balances to support the assessment of factor treatment effects, and modelled opportunities to improve organic matter supply to soils.

#### *2.1. Experimental Site and Trial Design*

The long-term field trial was initiated in 2010 Germany, Hesse (50◦11 39.0 N 8◦45 09.5 E) at 120 m above the sea level. It is maintained by the on-farm research and breeding department Dottenfelderhof. The soil type is a Haplic Luvisol with Silt loam from loess [23]. The average precipitation is 630 mm per year with an average temperature of 9.4 ◦C.

The farm was converted from conventional to biodynamic agriculture in 1968. In the time of conventional practice, sugar beet was cropped as a monoculture for many years. Since the conversion, the crop rotation consisted of a two times six year rotation with a legume/grass mixture in year one and two; winter wheat in year three; winter rye in year four; root crops in year five; and a spring cereal in year six. The legume/grass mixture alternated between clover/grass and alfalfa/grass from one six-year cycle to the next. Root crops varied widely and could be maize, potatoes, carrots, or other. The spring cereals are usually oats or spring wheat. In the rotation under study, it is important to notice that fodder maize was planted instead of winter rye in 2015 and clover/grass was ploughed and reseeded in 2013, because of drought and winter damage.

All treatments receive the same biodynamic preparations [24], i.e., BD 500 and BD 501 spray, at least once a year each. The compost used for the experiment was prepared with the usual biodynamic compost preparations and was made on site.

The trial was initiated in spring 2010 as a one factorial Latin square design with four treatments on plots of 48 m2 gross area (6 × 8 m) and 29.25 m<sup>2</sup> net area (4.5 × 6.5 m). On all plots, an equivalent livestock unit (LU) of 0.6 cattle deep litter (06M) was applied. Treatments 2 and 4 were treated with potassium sulfate (K), and treatment 3 and 4 with biodynamic compost (BD).


The cattle deep litter was a fermented manure from the farms' dairy cow herd. A total of 70% of the cow manure was distributed evenly, daily in the stable, and covered with straw. Cow pat pit preparation was added daily and compost preparations were applied once a month. The deep litter was harvested after the rye harvest and worked into the soil before the root crops were planted.

Potassium sulfate was produced by the fertilizer company K + S, under the tradename "Kalisop" and consisted of 50% water-soluble potassium oxide (K2O) and 45% watersoluble sulfur trioxide (SO3).

The biodynamic compost consisted of 85–90% green chop, 5–8% cow manure, and 5–7% soil. To speed up the process, the material was mixed daily in the first week and prepared with the cow pat pit during this time. After that, the single biodynamic compost preparations were added for the first time. Whey from the farm dairy or water was added to keep the right moisture content, which should be over 60% to avoid overheating and thus losses of nutrients, because the initial material was usually too dry. To protect the compost from rain, it was covered with a compost membrane. After the initial week and during the following half year process of composting, the compost pile was turned three to four times. After three months, the biodynamic compost preparations were added a second time.

Table 1 shows that the climatic water balance according to Haude [25] was negative from 2012 until 2015, and was positive in 2016 and 2017.

**Table 1.** Mean annual temperature, annual precipitation, and climatic water balance [25] during the investigation period.


#### *2.2. Fertilizer and Manure Application*

The applied amounts of manure and fertilizer are shown in Table 2, except an application of 2 Mg ha−<sup>1</sup> lime (CaCO3 with 56% CaO) on all treatments in November 2009, because the pH was too low at the start of the experiment. The cattle deep litter was applied on all treatments before planting of root crops once in a 6-year rotation.

**Table 2.** Crop rotation, amounts of organic amendment, and total nutrient amounts (kg ha<sup>−</sup>1) applied with all fertilizers (cattle deep litter, compost, and K2SO4).


<sup>1</sup> cattle manure from deep litter; <sup>2</sup> compost = plant-based compost; <sup>3</sup> K and S was applied at the same time as the compost in 2010, 2011, 2015, 2016, and 2017 as K2SO4.

The amount was calculated to represent 0.6 LU ha−<sup>1</sup> and was applied in spring 2010, before planting of potatoes (40 Mg ha−1) and in spring 2016 before planting of red beet (35 Mg ha−1). The same amount of compost (30 Mg ha−1) was applied on the 06M + BD and 06M + BD + K treatment in 2010 and from 2014 to 2017, after calculating the maximum allowed N amount by the German fertilizer regulation. In 2011, the applied amount of compost was 15 Mg ha−1. Potassium sulfate was applied on the 06M + K treatment in three subsequent years from 2015 to 2017, in an amount that was derived from previous dosing tests.

#### *2.3. Soil Samples and Chemical Analyses*

Soil samples were taken every year after harvest or in autumn, for clover grass, from a soil depth of 0–30 cm. These were then mixed and sent to the laboratory "Hessisches Landeslabor" (LHL).

Soil organic carbon (SOC) were analyzed by combustion at 550 ◦C under O2, using Leco® RC612 carbon analyzer. Total N were measured by the dry combustion method until 2012, according to DIN ISO 13878 [26], and afterwards according to DIN EN 16168 [27]. Total K, S, and P were determined by inductively coupled plasma optical emission spectrometry [28]. Soil pH was measured 1:10 in 0.01 M CaCl2 [29]. Soil bulk density was calculated as the dry weight of soil divided by its volume and as a mean of replications at the end of the rotation [30].

#### *2.4. Yield and Samples for Crop Nutrients*

Clover grass was cut three times during the vegetation period at 12 June, 1 August, and 10 October 2012, and two times in 2013 at 19 June and 24 September. The harvest from the net plots was weighed to determine fresh matter yield. A 5 kg mixed sample of harvest was chopped and from this material 2 × 1 kg was dried at 105 ◦C in an oven, to determine dry weight yield. Samples for the analyses of nutrient content were taken from the chopped material.

In 2014, winter wheat cv. Butaro was harvested with a Hege 125 combine. Grain and straw were weighed separately for fresh matter yield. The straw was processed in an analogous manner to clover grass, for determination of dry weight and laboratory samples.

From maize cv. Colisee in 2015, grain was harvested on 9 September by hand. The straw was harvested one day later with a maize chopper. Maize straw was processed in an analogous manner to clover grass. Red beet cv. Robuschka was harvested on 14 September by hand. Stem and leaves were separated from the bulbs, and fresh matter yield was determined separately. From both portions, a mixed sample of 2 kg was taken and sent to the laboratory. Sping wheat cv. Heliaro was harvested on 4 August and processed in an analogous manner to winter wheat.

Crop nutrients (P, K, and S) were measured with X-ray fluorescence spectroscopy, according to VDLUFA Volume III [31]. Dry matter and N were determined according to ISO 12099 [32].

#### *2.5. Soil Surface Nutrient Balance*

The nutrient balances were calculated from 2012 until 2017, because this was a full cycle of the crop rotation, beginning with the legume-grass mixture, until the spring cereal.

The N, P, K, and S balances were annually estimated as the difference between nutrient input and nutrient output (kg ha−<sup>1</sup> year<sup>−</sup>1):

$$\text{Nultrient budget} = \text{nutrient input} - \text{nutrient output} \tag{1}$$

Where the nutrient inputs included fertilization (deep litter manure, plant-based compost, and potassium sulfate) and crop seeds, the outputs included harvested aboveground biomass (main and side product).

For the N balance, the N inputs were extended by atmospheric N depositions, asymbiotic N fixation, and symbiotic N fixation. The N atmospheric deposition were estimated at 15 kg ha−<sup>1</sup> year<sup>−</sup>1, and the asymbiotic nitrogen fixation were 5 kg ha−<sup>1</sup> year<sup>−</sup>1.

The symbiotic N fixation was estimated according to the Stein-Bachinger [33]:

$$\text{Symbolotic N fixation} = (\text{N}\_{\text{shot}} + \text{N}\_{\text{root}} + \text{stubble}) \times \text{Leg}\_{\text{share}} \times \text{Ndfa} \tag{2}$$

where Nshoot was calculated as the product of grass-clover biomass and the N concentrations. Nroot + stubble was calculated as the product of grass-clover biomass and the fix value of 0.75 for the root and stubble biomass, and the totally fixed root and stubble N (1.5%). For the Legshare, we assumed the fix value 0.7 and for the Ndfa (nitrogen derived from the atmosphere) it was 0.8, respectively.

#### *2.6. Soil Organic Matter Balance (HU-MOD)*

The HU-MOD model [34,35] was developed as a decision support tool for application in farming practice. Unlike most other so-called humus balance methods, this model was conceptually able to analyze and predict soil organic matter changes [36]. The estimation of soil organic matter changes was based on the calculation of a coupled C and N balance in the soil–plant system. In principle, the model assumed that N in plant biomass could be used as a proxy for soil organic matter mineralization, if the N was supplied from other sources (here, atmospheric deposition, fertilizers, and—for legumes—biological nitrogen fixation) were considered. Thus, soil organic matter loss was calculated according to:

$$\text{NSOMIOSS} \left(\text{kg SOM-N ha}^{-1}\right) = \text{NPB} - \text{NFIX} - \text{NDEP} - \text{NFIZ} \tag{3}$$

NPB = N in total plant biomass (including roots), NFIX = N from biological fixation (legumes only), NDEP = N from atmospheric deposition, and NFTLZ = N from organic and mineral fertilizers.

SOM-N was transferred to SOM-C, based on the C:N ratio of the soil under assessment. Regarding the formation of new soil organic matter, the model applied a stoichiometric assumption, where the build-up of soil organic matter could be limited both by C and N availability. Again, the C:N ratio of the soil at the site under assessment was taken as a reference. Soil organic matter gain was therefore calculated according to:

$$\text{SOMGAIN} \left(\text{kg SOM-C ha}^{-1}\right) = \text{MIN} \left(\text{CREM; NREM} \times \text{SITECN}\right) \tag{4}$$

CREM = C from organic material (including plant roots), NREM = remaining N in the soil from organic material (including plat roots) and other inputs after consideration of losses, SITECN = reference C:N ratio of the soil at the site under assessment (topsoil C:N ratio was used as a proxy).

In the calculation of remaining C and N for the soil organic matter build-up, organic C and N inputs as well as mineral N inputs were considered. Losses of N in turnover were accounted for.

The model was successfully evaluated in several long-term and even in short-term field experiments [34,35,37].

#### *2.7. Statistical Analysis*

Data were analyzed using analysis of variance (ANOVA) for a Latin square design using SAS® Studio 3.8. Data normality was tested using the Shapiro–Wilk test (*p* < 0.05). Tukey's honestly significance difference (HSD) was used as a post-hoc mean separation test (*p* < 0.05), where the ANOVA performed significant. N stocks of 2014, 2016, and 2017 were reciprocally transformed.

#### **3. Results**

#### *3.1. Development of Soil Organic Matter and Nutrient Levels in the Soil*

In the course of the experiment, we observed an oscillating development of both carbon nitrogen stocks in soils, which were more pronounced with C (Figure 1a,b). The highest C values were measured in 2012 and 2017, which was at the start and the end of the first regular crop rotation. With N, the highest values were measured in 2010 and 2016/2017. In 2014–2016, the treatments with additional application of plant-based compost (06M + BD and 06M + BD + K) showed significantly higher stocks of SOC as compared to treatments without compost application (06M and 06M + K). The application of plant-based compost also led to a significant differentiation in the soil N stocks between the treatments in 2011 and after 2016. Nevertheless, soil total N stocks decreased from 2009 to 2017 by 17.7% and 12% for 06M and 06M + K, respectively. The 06M + BD and 06M + BD + K treatments maintained the initial values.

**Figure 1.** (**a**) Evolution of soil organic carbon stocks (Mg ha−1), (**b**) soil total N stocks (Mg ha−1), (**c**) soil total K stocks (Mg ha−1), (**d**) soil total S stocks (Mg ha−1), (**e**) soil total P stocks (Mg ha−1), and soil pH (**f**) in the soil layer of 0–30 cm, over the period of 2009–2017, as affected by different fertilization treatments. Error bars represent the standard error of the mean value. Different letters within a year are significantly different at *p* < 0.05.

Potassium (K) stocks were also oscillating, but the pattern was different from that of C and N (Figure 1c). In 06M + K, the highest values were measured after the potassium sulfate fertilization events in 2011 and 2014 (cf. Table 2). In 06M + BD + K, however, these events could not be identified. As expected, potassium sulfate fertilization with and without compost application led to significantly higher soil total K stocks. As compared to

06M, all other treatments maintained or increased K stocks by 3% and 5% for 06M + K and 06M + BD + K, respectively.

Sulfur (S) stocks were higher in 06M + K and 06M + BD + K after potassium sulfate application in 2010, 2015, and 2017, but not in the other years with additional K and S fertilization in these treatments (Figure 1d). The highest increase in S stocks was observed for 06M + BD + K (+ 38.8%), followed by 06M + K (+ 26.5%) and 06M + BD (+ 20%).

Moreover, the 06M + BD + K treatment resulted in a significant increase in the soil total P stock (2.61 Mg ha<sup>−</sup>1), as compared to the 06M (2.38 Mg ha−1) and 06M+K (2.38 Mg ha−1) treatments (Figure 1e), whereas 06M + BD were not significantly different from other treatments.

After 9 years, the soil pH in 06M, 06M + K, 06M + BD, and 06M + BD + K treatments were 0.3, 0.5, 0.6, and 0.4 units higher than the initial value in 2009 (Figure 1f). However, there were no statistically significant differences between the treatments in the last two years of the crop rotation.

#### *3.2. Yields over the Crop Rotation 2012–2017*

Depending on the crop and the year of investigation, the results varied. However, the different fertilization influenced the annual marketable yields, as shown in Table 3.

**Table 3.** Yields (Mg ha−<sup>1</sup> dry matter), nitrogen yields (kg ha−<sup>1</sup> dry matter), soil nitrate–nitrogen (mg kg−<sup>1</sup> 0–90 cm) in spring of the rotation. Means followed by different letters within a row are significantly different at *p* < 0.05.


Abbreviations: SEM, standard error of the mean value; and LSD, Least Significant Difference.

The yields differed significantly in 2012, 2014, 2016, and 2017, between 06M and 06M + BD + K. Further, the application of compost plus potassium sulfate resulted in higher N yields of all treatments, as compared to 06M in all years, except for 2013. Despite the different N input, the mineral N in spring was similar in all treatments.

Fertilization resulted in significant marketable yield increases cumulated over the 6-year crop rotation, which followed the order—06M < 06M + K < 06M + BD < 06M + BD + K (Figure 2a). The significantly highest marketable yields cumulated over the 6-year crop rotation was achieved with the addition of plant-based compost, with and without potassium sulfate (26.1 Mg ha−<sup>1</sup> and 25.4 Mg ha<sup>−</sup>1, respectively), while the treatment with only deep litter (06M) achieved 22.7 Mg ha<sup>−</sup>1, which were 13% and 10.6% less than 06M + BD + K and 06M + BD.

**Figure 2.** (**a**) Marketable yields (Mg ha−<sup>1</sup> dry matter) included winter wheat, maize, red beet, and spring wheat. (**b**) Total nitrogen yields (kg ha−<sup>1</sup> dry matter) included clover grass, winter wheat (grain and straw), maize, red beet (root and side product), and spring wheat (grain and straw). Results are cumulated over the 6-year crop rotation, boxplots with different letters are significantly different at *p* < 0.05.

Fertilization with potassium sulfate significantly influenced the total aboveground biomass N uptake over the crop rotation, being significantly higher in 06M + K and 06M + BD + K than in 06M (Figure 2b). Compost application (06M + BD) did not increase the N yield as compared to 06M and 06M + K, but was significantly lower than the combination of compost and potassium sulfate.

#### *3.3. Nutrient Balance over the Crop Rotation*

Nutrient inputs in the treatments varied according to the fertilization regimes, and the exports varied according to the yield levels. The N:P:K:S ratios were only different between treatments on the input side, but not for the nutrient exports.

Balances of all nutrients under study were negative with the 06M treatment (Table 4). Inputs did not compensate for nutrient export in this treatment.



Potassium sulfate application turned the K and S balances positive in the 06M + K treatment, while the P and N budgets became positive only with compost application in the experiment (treatments 06M + BD and 06M + BD + K).

#### *3.4. Soil Organic Matter Balances and Modeling*

The good correlation between observed and predicted C and N development (Figure 3) indicated that assumptions in the model seemed to be more or less applicable at the site, even though the undulating development of SOC was not captured in that magnitude by the model.

**Figure 3.** Observed and predicted development of soil organic C (**a**) and N (**b**) under the rotational cycle 2012–2017 in the long-term field experiment. Predicted values were calculated with the HU-MOD model.

According to the coupled C- and N-based soil organic matter balance, the supply of organic matter was too low in the 06M and 06M + K treatments to compensate for mineralization (Table 5). With compost application in 06M + BD and 06M + BD + K, the balance became slightly positive, but there was not much potential for increasing yields.

Modeling opportunities to improve organic matter supply to soils in treatments with and without compost (Table 6), we found that the inclusion of non-legume catch crops would marginally improve SOM balances, but the budgets would almost not change in 06M. An optimization of the crop rotation (substitution of fodder maize by oats and new crop order) would significantly improve the SOM budget, but still the balance of the 06M treatment would stay negative. In the compost treatments, the same optimization of the crop rotation would allow for a 50% reduction of compost application, without significantly changing the budget (Table 7).

**Table 5.** Coupled C- and N-based soil organic matter balance with HU-MOD across one crop rotation (2012–2017) in kg ha−<sup>1</sup> year<sup>−</sup>1.


**Table 6.** Scenarios for optimized crop rotation for treatment by 06M. Coupled C- and N-based soil organic matter balance with HU-MOD across one crop rotation (2012–2017) in kg ha−<sup>1</sup> year<sup>−</sup>1. For the scenarios, the CN reference was standardized, therefore, the budgets of the original scenarios differed from those in Table 5.


\* Fertilization: 35 Mg ha−<sup>1</sup> with farmyard manure # Catch crop effect is considered in the balance value of the main crop.

**Table 7.** Scenario for the optimized crop rotation for treatment by 06M + BD + K. Coupled C- and N-based soil organic matter balance with the HU-MOD across one crop rotation (2012–2017) in kg ha−<sup>1</sup> year−1. For the scenario, the CN reference was standardized, therefore, the budgets of the original scenario differed from those in Table 5.


\* FYM = farmyard manure; # Catch crop effect is considered in the balance value of the main crop.

#### **4. Discussion**

#### *4.1. Soil Organic Matter and Nutrients*

The development of soil organic matter is a rather slow process, and it will likely take more than a 6-year crop rotation or even an 8-year observation period, until significant treatment effects emerge [38]. However, we can already observe some differentiation in this initial phase of the field experiment. As could be expected, both C and N values were higher in the compost treatments than in the other two treatments. Further, it appears that decreasing C and N stocks in the treatments without compost application could indicate an insufficient supply of organic matter to soils. However, since the turnover of organic matter in soils is not only dependent on actual management, but also on site conditions and management history [36,39,40], it is not possible to determine whether farmyard manure application corresponding to a stocking rate of 0.6 LU cattle ha−<sup>1</sup> is insufficient in general, or only worked under the specific conditions of this experiment.

Unfortunately, there are almost no studies on the stocking rate and the corresponding available manure effects on soil fertility in the scientific literature today. In the well-known DOK experiments in Switzerland, farmyard manure was applied at rates corresponding to 0.7 and 1.4 LU cattle per ha, but soil carbon stocks decreased under all treatments in the experiment, except for a biodynamic treatment with composted manure application, corresponding to 1.4 LU ha−<sup>1</sup> [41]. This was most likely an effect of the site history. Leithold et al. [3] assumed that 1 LU ha−<sup>1</sup> would be an adequate stocking rate to maintain soil fertility in productive organic farming systems. Nevertheless, Schulz et al. [4] observed even increasing SOM levels in the Organic Arable Farming Experiment Gladbacherhof (Germany) at the FYM application, corresponding to 1 LU cattle per ha.

Soil organic matter balances provide some explanation for the observed trends of soil C and N, despite the uncertainties in the parametrization (see below). According to the model calculations, organic matter supply is not sufficient in the treatments without compost application to compensate for turnover losses, despite the low yield level. It was considered that the N uptake of crops is taken as a proxy for soil organic matter mineralization in the model [34]. The yield level of non-legume crops is, therefore, positively correlated with the demand for organic matter to compensate for SOM mineralization [42]. At present, only treatments with compost applications have the potential to build up SOM. However, the demand for organic matter would increase if the yield levels shall be improved in the experiment.

The development of K, S, and P stocks reflects the different input rates in the treatments. Therefore, treatments with compost application (06M + BD, 06M + BD + K) have higher p-values than the two other treatments, and treatments with potassium sulfate application (06M + K, 06M + BD + K) have higher K and S values.

#### *4.2. Crop Yields*

The fodder legumes obviously benefitted from potassium sulfate application. From our results we might not conclude whether this was a K- or S-effect. Both elements play a vital role in biological nitrogen fixation, and the effect of variable availability is similar [17]. Usually, K is not a limiting factor in arable soils in Germany. At the site of the field experiment, however, K supply might be limited by K fixation.

Row crops (maize and red beet) both benefitted from compost application, while K/S fertilization had a smaller effect. Here, it must be considered that all treatments received farmyard manure. According to Blake et al. [43], farmyard manure application is more effective in supplying crops with K than mineral fertilizers. Lehtinen et al. [10] found that K input with manure-based compost was higher than with plant-based compost. In fact, we found that K use efficiency was highest in the 06M treatment and lowest in the 06M + BD + K.

The reaction of the cereals to fertilization was not consistent. Higher winter wheat yields in the treatments with K fertilization are likely an effect of the preceding fodder legumes. Spring wheat at the end of the crop rotation on the other hand obviously profited from compost fertilization.

Altogether, yield levels were comparably low in the experiment. Nutrient balances revealed that the actual nutrient supply did not offer much potential for yield increases, if at all. Increasing yields would require additional efforts in soil fertility management, like the use of green manure and catch crops to improve N supply [44], and additional fertilization.

#### *4.3. Nutrient and Soil Organic Matter Balances*

Nutrient balances were negative for both S, P, and K in the control treatment (06M). This corresponded to the results of Berry et al. [45], which indicate negative K and P balances in organic farming systems without external inputs. In our field experiment, the application of compost compensated for nutrient exports (and losses), even in the treatment without additional Potassium sulfate fertilization. The utilization of additional internal organic matter resources, therefore, is a good opportunity to improve nutrient balances on arable land, but of course this measure did not close the nutrient cycle on the farm level. Instead, nutrients were transferred and re-distributed within the farm. Reimer et al. [46] addressed this situation in their meta-analysis of nutrient budgets in organic farms in Germany. The authors emphasized that positive nutrient balances at the farm gate could not be achieved without nutrient imports. In principle, biowaste compost and sewage sludge would be the appropriate sources to close nutrient cycles. However, both sources featured the risk to import mineral and organic pollutants into organic farming systems [47]. Therefore, the utilization of internal resources must be considered a viable interim option.

Soil organic matter balances were calculated with the HU-MOD model [34,35]. The tool was originally developed for decision support in farming practice, but unlike most other so-called 'Humus balance methods' it could also be used for analytical purposes [36].

The advantage of the HU-MOD model was that the utilization of N in plant biomass as a proxy for the mineralization of soil organic matter allowed us to by-pass the need for information on site factors, as it was assumed that their effect on soil organic matter mineralization became visible in the N fluxes that were considered in the model. However, the procedure made the model susceptible to errors resulting from erroneous estimates of N pools. Most importantly, biological nitrogen fixation is known to be highly variable [48], even though [49] found that there is a statistically significant average rate of approximately 0.7 kg BNF-N per kg plant shoot N. Nevertheless, the error of this average was high enough to severely impact site-specific N balance calculations. As we did not measure BNF, we could not account for any differentiation between the treatments in N yield from this process.

Regarding the congruence between the observed and predicted trends of soil C and N stocks, it must be considered that the model output does refers to the total rooted soil layer. A comparison with topsoil C and N trends, therefore, comprises the risk that C and N changes in deeper soil layers are not captured. Soil organic matter in the subsoil is usually more stable than topsoil SOM [50], and turnover mainly takes place in the topsoil. However, it was argued that organic matter turnover in subsoil is relevant for the calculation of actual C balances [51]. Therefore, it cannot be concluded whether the deviations between measured topsoil organic matter changes and the calculated changes according to the HU-MOD model indicate parametrization errors, or are caused by the limited database on C and N changes in the soil.

#### *4.4. Practical Implications*

#### 4.4.1. Farm Compost to Open Additional Organic Matter Sources

In our experiment, we used on-farm composting as an option to increase organic matter supply to soils based on own resources of a farm. This compost provided an additional input of several nutrients. Besides N, the compost contained considerable amounts of P, K, and even S and could be considered an effective 'full-fertilizer' with a low leaching potential [52]. Compost application therefore could effectively increase crop yields [11,13,53].

As outlined by D'Hose et al. [53], the positive effect of farm compost on crop yields should not only be ascribed to the additional nutrient input, but also to the improvement of growing conditions for the crops. Compost application improves soil physical properties [52,54], and has a beneficial effect of compost on pathogen regulation in soils and plant health [54–56]. Further, compost might even facilitate the formation of arbuscular mycorrhizas [57].

It is widely acknowledged that compost builds up soil organic matter [8,10,52,58]. This improves microbial activity and related ecosystem services [59]. For example, organic matter build-up improves the accessibility of micronutrients to plants [14]. In general, there is a positive relationship between soil organic matter and crop yields [15,42].

Composting was also identified as a viable option to reduce ecological trade-offs between soil fertility management and environment and climate protection [60,61].

#### 4.4.2. Optimization

Modeling possible adaptations of crop rotation or fertilization in selected treatments indicated a low potential of catch crops to improve the SOM balance. This is supported by findings of White et al. [58] as well as Tautges et al. [12]. However, it contradicts the results of Poeplau and Don [62] who concluded that the introduction of catch crops into crop rotations would have a considerable effect on carbon sequestration in Germany. The reason for the different observations could be a stoichiometric effect, were N availability limits C retention from catch crops in the soil [63]. As the C:N ratio of organic material is narrowed down in the turnover process, excess C is lost by respiration [64]. At the same time, it should be considered that increasing N supply causes a priming effect [65], which pushes organic matter turnover. The impact of catch crops on soil organic matter is therefore probably dependent on both C and N amounts and availability, alongside with biological and physical factors. As we included non-legume catch crops in the model that did not receive any fertilization, the model calculated only slightly positive balances based on the fertilization effect of the catch crops on the succeeding crops. However, it should be considered that soil N taken up by the catch crops might have leached in a corresponding bare fallow period. On the other hand, N leaching usually is very low under the N-limited conditions of organic farming [66], especially on heavy soils.

In contrast, the substitution of fodder maize by oats proved a very effective measure in the modeling study, as this adaptation significantly decreased N export and the related demand for organic matter. Field experiments comparing maize and cereal cultivation effects on soil organic matter are rare. Nevertheless, in a study from Poland, Rychcik et al. [67] found that maize and grain legumes had lower soil carbon values than cereals. In a recent paper, Benbi et al. [68] showed that soil C respiration was three-fold higher under maize, as compared to wheat. Our results therefore are plausible.

Of course, changes in the crop rotation need to be discussed against the background of the requirements in the farming system. A substitution of fodder maize by cereals might not always fit to the specific situation. In such cases, intercropping could provide an option to improve the soil organic matter balance of maize [69].

#### **5. Conclusions**

Sustainable nutrient supply might be threatened in organic farming systems with a stocking rate of 0.6 LU cattle per hectare, if fertilization only relies on the available manure. Additional compost application provides a solution, as compost provides a direct additional input of nutrients, and contributes to the nutrition of legumes, which in turn enhances biological N fixation. Additional supply of essential nutrients (K, S) does further improve BNF. This compost can be made from internal resources on the farm (e.g., hedgecutting), to be independent from external inputs and to avoid the import of pollutants. However, it must be considered that own farm compost makes new nutrient, and the organic matter sources available on the farm and is a viable interim solution, but does not solve the problem of open nutrient cycles at the farm gate level.

**Author Contributions:** Conceptualization, M.O., C.B., and B.S.; methodology, B.S. and M.O.; validation, H.S. (Hartmut Spieß) and C.B.; formal analysis, M.O.; investigation, C.M., H.S. (Harald Schaaf), and D.B.; resources, H.S. (Harald Schaaf), D.B. and H.H.; data curation, C.M., B.S., and M.O.; writing—original draft preparation, M.O. and B.S.; writing—review and editing, C.B. and H.S. (Hartmut Spieß); visualization, M.O.; supervision, C.B.; project administration, C.B.; funding acquisition, C.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Software AG-Stiftung, Rudolf-Steiner-Fonds and Zukunftsstiftung Landwirtschaft. The APC was funded by Software AG-Stiftung.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** All data are available at Forschungsring and Forschung&Züchtung Dottenfelderhof.

**Acknowledgments:** The authors acknowledge the support of member of the advisory board of the BoDyn Long-Term Field Experiment (Miriam Athmann, Johan Bachinger, Andreas Gattinger, Jürgen Fritz, Ulrich Köpke, Harald Schmidt, Klaus Wais) and of Cornelius Sträßer, the responsible project supervisor of Software AG-Stiftung.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Effect of Drip Fertigation with Nitrogen on Yield and Nutritive Value of Melon Cultivated on a Very Light Soil**

**Roman Rolbiecki 1, Stanisław Rolbiecki 1, Anna Figas 2, Barbara Jagosz 3, Dorota Wichrowska 4, Wiesław Ptach 5, Piotr Prus 6,\*, Hicran A. Sadan 1, Pal-Fam Ferenc 7, Piotr Stachowski <sup>8</sup> and Daniel Liberacki <sup>8</sup>**


**Abstract:** Most species of Cucurbitaceae respond favorably to irrigation, especially when combined with fertilizers. The effect of drip irrigation combined with nitrogen fertigation in melon grown on a very light soil in Central Poland, during 2013–2015, was evaluated. The field experimental design was a split-plot with four replications. Two factors were studied: (1) irrigation treatments applied in two combinations—drip irrigation + broadcast nitrogen fertilization (control), and drip irrigation + fertigation with nitrogen; (2) two cultivars—Melba and Seledyn. The total marketable yield of fruits, weight of a single fruit, and the concentration of dry matter, total sugars, monosaccharides, ascorbic acid, total carotenoids, and polyphenols were evaluated. Tested factors presented a significant effect both on the yield and nutritive value characteristics. Drip irrigation combined with nitrogen fertigation, comparing to the control, notably improved yields and nutritional value of fruits. Seledyn produced better yields than Melba. This study shows that on very light soil, with low water and nutrient retention capacity, melon should be drip-irrigated and nitrogen-fertigated to obtain the best cultivation results.

**Keywords:** *Cucumis melo* L.; chemical composition; cultivar; drip irrigation; fruit quality

#### **1. Introduction**

Melon (*Cucumis melo* L.) belongs taxonomically to the Cucurbitaceae family, which also includes vegetables, such as cucumber, pumpkin, squash, watermelon, and gourds. In many countries around the world, melon fruit is of considerable economic importance. World production of this species in 2018 was estimated at 40 million tons per year. The main melon-producing country is China (12.7 million tons per year), followed by Turkey, Iran, and India (1.8 to 1.2 million tons per year) [1].

There are many cultivars of melon, which differ mainly in shape, color, and taste [2]. Melon fruits are valuable in terms of nutritional and bioactive properties. This species is a

**Citation:** Rolbiecki, R.; Rolbiecki, S.; Figas, A.; Jagosz, B.; Wichrowska, D.; Ptach, W.; Prus, P.; Sadan, H.A.; Ferenc, P.-F.; Stachowski, P.; et al. Effect of Drip Fertigation with Nitrogen on Yield and Nutritive Value of Melon Cultivated on a Very Light Soil. *Agronomy* **2021**, *11*, 934. https://doi.org/10.3390/ agronomy11050934

Academic Editors: Nikolaos Monokrousos and Efimia M. Papatheodorou

Received: 26 March 2021 Accepted: 5 May 2021 Published: 9 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

very good source of carotenoids (α-, β-carotene, and β-cryptoxanthin), folic acid, pectins, as well as many vitamins (including B group) and minerals (mostly potassium, iron, and magnesium), polyphenols, such as flavonoids and phenolic acids, and fatty acids (including oleic, linoleic, and palmitoleic acids). Melon is a fruit appreciated not only for its taste and dietary qualities, but also its healing properties, thanks to which it is also used in the cosmetics industry [3–6].

In Poland, due to unfavorable climatic conditions, the melon is grown as a noncommercial species. However, the interest in melon cultivation is clearly growing every year. Currently, the Polish National List of Vegetable Plant Varieties includes seven cultivars that are suitable for cultivation in Poland, and their number is systematically growing [7]. This species is photophilous and thermophilic, with a very high water requirement. Melon plants are very sensitive to spring and autumn frosts, which negatively affect growth and development, and thus also the fruit yield [8]. The highest sensitivity of melon plants to water deficit is observed during the fruit setting period [9].

Due to the rising interest in melon growing in Poland, it is necessary to broaden the knowledge about the methods of its cultivation in temperate climatic conditions. Field production of melon largely depends on the thermal conditions and precipitation during the growing season. An important factor in obtaining high- and good-quality crops is ensuring optimal soil moisture during the vegetation period of this species. In Poland, the water requirements of plants from the Cucurbitaceae family are estimated at around 400 mm during the growing season. The main reason for the high water needs of plants belonging to this family is their high fertility and the production of much aboveground mass with a high coefficient of transpiration (as the ratio of the amount of water excreted to the production of dry matter) [10]. It is generally accepted that irrigation significantly affects both the melon yield and the components of the melon yield grown under semiarid climatic conditions [9,11–13]. Many studies have shown that the field cultivation of melon should be carried out using irrigation treatments [11–16]. It was found that production factors such as water and nutrients (nitrogen, phosphorus, potassium) most often limit the possibility of obtaining a higher yield of melon fruit [17,18]. Drip irrigation combined with fertigation is a good way to increase the efficiency of water use and yield of Cucurbitaceae. It was also found that drip irrigation performed during the cultivation of Cucurbitaceae and other vegetables sensitive to climatic conditions during the growing season clearly increases their nutrient concentration [19,20]. Drip fertigation ensures precise administration of appropriate amounts of nutrients directly to the root zone. Accurate and uniform application of macro- and micronutrients adequately meets the needs of crops during the growing season [21–23].

The objective of this study was to evaluate the effect of drip irrigation combined with nitrogen fertigation on the melon fruit yield and nutritive value characteristics. As a control, drip irrigation combined with broadcast fertilization was used. The total marketable yield of fruits, weight of a single fruit, and the concentration of dry matter, total sugars, monosaccharides, ascorbic acid, total carotenoids, and polyphenols of two melon cultivars (Melba and Seledyn) were evaluated. The experiment was carried out on very light soil in a region of high precipitation deficit; hence, the advisability of irrigation treatments in this area is justified.

#### **2. Materials and Methods**

#### *2.1. Field Experiment Description*

The field experiment involving the drip fertigation of two melon (*Cucumis melo* L.) cultivars, namely Melba and Seledyn, was conducted in Kruszyn Kraje ´nski near Bydgoszcz 53◦04 53 N, 17◦51 52 E (Central Poland). The area has precipitation deficits, an extremely unfavorable water balance, and high frequency of long periods without rainfall [19,22–29]. The study was carried out in the years 2013–2015. The plants were grown using standard crop management practices recommended for melon cultivation in Poland. The study was carried out on very light soil with a weak and very weak rye–soil complex. Based on the

percentage content of individual granulometric fractions, this soil was classified as sand [30]. The soil of the experimental field contained such fractions as: sand—86.97% (from 2.0 mm to 0.05 mm), silt—12.28% (from 0.05 mm to 0.002 mm), and clay 0.75% (<0.002 mm). The average content of total organic carbon and concentration total nitrogen in the soil was 9.6 g kg−<sup>1</sup> and 0.9 g kg−1, respectively. The experimental soil was characterized by a low capacity for water retention. The water reserve to 0.6 m depth of soil at field capacity was 72.7 mm, at wilting point 29.1 mm, and the available water 43.6 mm.

The experiment was conducted as a split-plot design with four replications. Two factors were used in the study. The first factor was the drip fertigation with nitrogen applied in two combinations: (1) drip irrigation + broadcast nitrogen fertilization (control); (2) drip irrigation + fertigation with nitrogen. The second factor was two melon cultivars: Melba and Seledyn.

Melon seedlings were transplanted at 0.6 m within rows and 1.6 m between rows. The area of each harvest plot was 12 m<sup>2</sup> and included 15 melon plants, and the whole experimental plot size was 274 m2. Before planting the seedlings, cultivating and harrowing were performed. The fertilization consisted of 120:100:150 kg ha−<sup>1</sup> of nitrogen: phosphorus: potassium. The fertilization of phosphorus and potassium was carried out every year in early spring. The doses of potassium (potash salt) and phosphorus (superphoshate) fertilization depended on the abundance of these nutrients in the soil, based on the soil analysis carried out each year. Nitrogen fertilization (ammonium nitrate) was applied in three doses of 40 kg N ha−<sup>1</sup> during the growing season for both variants of fertilization. Fertigation was carried out using a proportional fertilizer dispenser. Drip irrigation and drip fertigation were carried out using the "T-Tape" drip line with a distance of 20 cm between the emitters. The efficiency of a single emitter was 1 l h<sup>−</sup>1. The distance between the drip lines was 1.6 m. Water from the subsurface well was used for irrigation. The quality and physical and chemical properties of the irrigation water used complied with the quality standards for irrigation water. Drip irrigation was started when the water potential in the soil was close to –40 kPa and finished when the water potential in the soil was close to –10 kPa. The end of irrigation treatments was determined on the basis of soil water potential at field water capacity, measured with a tensiometer. The tensiometers have been installed at every variant of the experiment at the depth of 25 cm. The dates of planting during the particular growing seasons were in the second week of June. Harvesting took place at the physiological stage of fruit ripeness (from the beginning of 3rd week of August till 1st week of September). Ripe fruits were picked progressively as they matured. In the experiment, the total marketable yield of melon fruits (t ha<sup>−</sup>1) and weight of single melon fruit (kg) were assessed.

#### *2.2. Nutritive Value Assessment*

To carry out a nutritional assessment the fresh melon fruits, one fruit from all plants in one plot was cut into a 5 cm wedge and then cut into 1-cm-thick slices. The frozen material was lyophilized (model Alpha 1–4 LDplus, Donserv, Warszawa, Poland), in order to achieve a permanent weight, and then it was ground to a fine powder (the particles were 0.3–0.5 mm in size) and was milled using the ultracentrifuge (Model FW177, Chemland, Stargard, Poland). The ground samples were stored in the dark, in bags, which were placed in desiccators for further analysis.

The total dry matter content of 'Melba' and 'Seledyn' melon fruits was determined using the drying technique according to the methodology of the Association of Official Analytical Chemists [31].

Carbohydrate analyses were performed according to Talburt and Smith's [32] procedures. For reducing sugar concentration assessment, one gram of freeze-dried material sample was placed in a 250 mL bottle; 150 mL of distilled water was then added and it was shaken vigorously. One milliliter of the filtrate was mixed with 3 mL of DNP reagent in a test tube and then heated in a water bath at 95 ◦C for 6 min. Absorbance of the mixture was measured using a spectrophotometer at a wavelength of 600 nm. The reducing

sugar concentration was then estimated using the standard curve of glucose. The total soluble carbohydrate was determined after hydrolysis of sugars. After filtration, 40 mL of the filtrate was taken, and 2 drops of concentrated HCl were added. The samples were warmed in a water bath for 30 min. After cooling, the mixture was neutralized using concentrated NaOH until pH 8.0 was reached. Next, 1 mL of the filtrate was mixed with 3 mL of DNP reagent and the procedure for determining the concentration of reducing sugars was followed. The results were converted to fresh weight taking into account the percentage of dry weight in the fresh matter.

Ascorbic acid reducing sugar concentration was assessed according to Kapur et al. [33]. Ten grams of fresh melon sample was homogenized with 25 mL of 2% oxalic acid solution and quantitatively transferred into a 50 mL volumetric flask and shaken gently to homogenize the solution. Then, it was diluted up to the mark with oxalic acid solution. The obtained solution was then filtered and centrifuged at 4000 rpm for 15 min, after which the supernatant solution was used for spectrophotometric determination (UV–1800, UV Spectrophotometer System, Shimadzu, Kyoto, Japan) of ascorbic acid concentration. Ascorbic acid is oxidized to dehydroascorbic acid by adding bromine water. After this, L—dehydroascorbic acid reacts with 2,4—DNPH and produces an osazone, which, treated with 85% H2SO4, forms a red-colored solution. A typical calibration plot was made and used to determine the concentration of ascorbic acid in the investigated samples.

Total carotenoids in melon samples were extracted by procedures described by Herrero-Martinez et al. [34]. Ten grams of lyophilized melon was blended with 100 mL saturated anhydrous sodium carbonate and mixed with a mechanical blender. Ten grams of the mixture was transferred into a centrifuge tube, 20 mL tetrahydrofuran was added, and it was mixed for 2 min under cold water. The mixture was centrifuged at 5000 rpm for 5 min and the supernatant was collected. Extraction was performed by adding 15 mL dichloromethane and 15 mL of 10% w/v NaCl into the supernatant and shaking it for 2 min. The extraction was repeated twice; the organic layer was collected and evaporated under nitrogen steam. The residue was kept at –20 ◦C, reconstituted with 5 mL dichloromethane, and diluted (1/40-fold) with dichloromethane prior UV measurements (Shimadzu UV-1800, UV–Vis spectral photometer system, Japan). Detection was performed at 450 nm according to the procedure reported in the Polish Standard [35]. Standard β-carotene for identification was prepared in dichloromethane to obtain 4 μg mL<sup>−</sup>1.

Total phenolic reducing sugar concentration was determined using the Folin–Ciocalteu reagent (Sigma-Aldrich, Darmstadt, Germany) according to the method of Singleton and Orthofer [36]. A volume of 0.5 mL of Folin–Ciocalteu reagent previously diluted with distilled water (1:10) was mixed with 0.1 mL of each sample. The solution was allowed to stand for 5 min at 25 ◦C before adding 1.7 mL of sodium carbonate solution (20%). Then, 10 mL of distilled water was added to the mixture, and the absorbance was measured at *λ* = 735 nm after 20 min of incubation with agitation at room temperature. Results were expressed in mg of gallic acid equivalents (GAE) per kg of fresh sample.

#### *2.3. Statistical Analysis*

All the experimental data were tested for differences by two-way ANOVA using of Statistica® 13.1 package. The significance of differences (LSD—lowest significant difference) was evaluated using the Tukey multiple confidence intervals for the significance level of *p* = 0.05.

#### *2.4. Weather Conditions*

The average air temperature in Kruszyn Kraje ´nski in the vegetation period, i.e., from 1 April to 30 September in the years 2013–2015, was 14.9 ◦C and was 0.3 ◦C higher than the mean for the long-term period 1986–2015 (Table 1). The warmest month of the growing season in 2013–2015 was July, with a mean temperature of 19.6 ◦C (0.8 ◦C above the mean for long-term period). In 2014, the highest average air temperature (15.4 ◦C) was recorded, which was 0.8 ◦C higher compared to the mean for the long-term period.


**Table 1.** Average air temperature (◦C) data during the vegetation period of Melba and Seledyn melon cultivars in the years 2013–2015.

The mean sum of precipitation in Kruszyn Kraje ´nski in the period from 1 April to 30 September, for the years 2013–2015, amounted to 279.2 mm and was 31.4 mm lower than the mean for the long-term period 1986–2015 (Table 2). The highest precipitation during the vegetation period occurred in 2013 and amounted to 354.3 mm (43.7 mm above the mean for the long-term period). In the 2015 growing season, the lowest total precipitation was recorded, amounting to 193.3 mm, and was 117.3 mm below the mean for the long-term period. The mean precipitation in April, June, July, and August in 2013–2015 was lower than the mean for the long-term period. The highest monthly precipitation (91.7 mm in May and 79.0 mm in July) was noted in the growing season of 2013.

**Table 2.** Precipitation (mm) data during the vegetation period of Melba and Seledyn melon cultivars in the years 2013–2015.


#### *2.5. Irrigation Water Rates*

The seasonal irrigation water rates used in the growing of Melba and Seledyn melon cultivars were inversely proportional to rainfall amount during the irrigation period. Relationship between precipitation (mm) and seasonal irrigation water rates (mm) of Melba and Seledyn melon cultivars in June–August in the years 2013–2015 is shown in Figure 1. The melon irrigation period, mean for 2013–2015, began on 13 June and ended on 4 August and lasted for an average of 53 days. The shortest irrigation period, only 11 days, was carried out in 2013. On average, in 2013–2015, during the irrigation period, 14 single waterings took place. The average seasonal dose, in the years 2013–2015, was 142.2 mm and ranged from 104.5 mm in 2013 to 169.0 mm in 2014. Both experimental treatments received the same amount of irrigation water.

**Figure 1.** Relationship between precipitation (mm) and seasonal irrigation water rates (mm) of Melba and Seledyn melon cultivars in June–August in the years 2013–2015.

#### **3. Results and Discussion**

In the control field (the drip irrigation combined with broadcast nitrogen fertilization), the mean total marketable melon yield in the years 2013–2015 was 26.38 t ha−<sup>1</sup> and 32.69 t ha−<sup>1</sup> for Melba and Seledyn cultivars, respectively (Table 3). On average, for the two studied melon cultivars, the highest marketable fruit yield was recorded in 2014. With the drip irrigation and broadcast nitrogen fertilization, the marketable yield was 34.95 t ha<sup>−</sup>1, and in the field with the drip irrigation and fertigation with nitrogen, the yield of fruits was 39.85 t ha−1. The lowest values of this parameter were recorded in 2015, when the marketable yield was 19.96 t ha−<sup>1</sup> and 22.68 t ha−1, respectively, for the control and for drip irrigation combined with nitrogen fertigation. Compared to the control, the liquid fertigation significantly increased the fruit yield of Melba and Seledyn cultivars by 12.5% and 13.6%, respectively. There was no significant interaction between irrigation treatment and cultivars.

**Table 3.** Influence of drip fertigation on the total marketable yield of fruits (t ha<sup>−</sup>1) of Melba and Seledyn melon cultivars in the years 2013–2015.


<sup>1</sup> LSD = the lowest significant difference (Tukey's confidence half-interval) at *p* < 0.05; n.s.—not significant at *p* < 0.05.

Some studies have previously confirmed the beneficial effect of irrigation combined with fertilization on the development of plants of the Cucurbitaceae family during cultivation on light soils. The marketable yield of watermelon grown on light soil in Central Poland under the influence of irrigation combined with nitrogen fertilization increased

by an average of 21% [23]. In addition, in the cultivation of watermelon performed in semi-arid regions of Brazil, on the sandy soil with low retention capacity of water and low nutrient levels, the authors reported a significant effect of irrigation and nitrogen fertigation on the plant growth, increasing the yield by 64% [37]. In the study presented in this paper, the marketable yield of melon fruits was comparable to the yields obtained in research performed in other soil and climate conditions in other regions of the world. In a study carried out in Turkey, the yield of melon ranged from 18.0 t ha−<sup>1</sup> to 32.4 t ha−<sup>1</sup> depending on the irrigation method [38], and from 8.2 t ha−<sup>1</sup> to 43.8 t ha−<sup>1</sup> depending on the year of research and the type of irrigation system [39]. In an experiment carried out under field conditions with furrow irrigation in Northwest China, near to the Tengger dessert, the yield of melon ranged from 19.6 t ha−<sup>1</sup> to 27.8 t ha−<sup>1</sup> [40]. In research performed in Northern Jordan, melon fruit yields ranged from 15.6 t ha−<sup>1</sup> to 23.5 t ha−<sup>1</sup> depending on irrigation quantity [41]. In other studies carried out in Turkey, different irrigation systems and nitrogen levels affected the fruit yield of melon [42]. In the above research, by analyzing different levels of melon irrigation and fertilization, positive effects of combining nitrogen fertilization at a dose 60 kg N ha−<sup>1</sup> and drip irrigation were observed. As a result of these experiments, the fruit yield of melon was 59.77 t ha<sup>−</sup>1.

In the present study, a significant influence of the cultivar on the marketable yield of melon fruit was also noticed. The marketable yield of fruits of the Seledyn cultivar was higher by 24.5% compared to the Melba cultivar. Significant relationships between the yield characteristics and the cultivar of watermelon have already been observed in previous studies, the purpose of which was to compare the effects of irrigation and fertilization on the fruit yield [23,43,44].

The increase in the marketable yield of melon fruits results primarily from a significant increase in the single fruit weight. The lowest total marketable yield of fruits and weight of a single fruit was obtained in 2015 (Tables 3 and 4). According to meteorological data, 2015 was very dry. The total rainfall in the period from April to September was only 193.3 mm (62% of the mean for the long-term period 1986–2015). The average air temperatures during the growing season in April, May, June, and July were lower than the mean for the long-term period, 1986–2015, by 0.6 ◦C, 0.9 ◦C, 0.6 ◦ C, and 0.3 ◦C, respectively (Table 1). Water deficits negatively affect the development of the melon, as it is a photophilous and thermophilic species, with a very high water requirement [8,9]. The melon plants of the Seledyn cultivar produced fruits of significantly greater weight than the plants of the Melba cultivar (Table 4). The use of drip fertigation with nitrogen significantly increased the average melon fruit weight of both the Melba cultivar by 0.12 kg (average fruit weight 0.76 kg) and the Seledyn cultivar by 0.19 kg (average fruit weight 1.15 kg). There was no significant interaction between irrigation treatment and cultivar. Melba is an early cultivar with an average weight of one fruit ranging from 0.5 kg to 0.7 kg [8]. In turn, the Seledyn cultivar is one of the very early ones with fruit larger than Melba, weighing up to 1.4 kg. For comparison, in the study carried out in Turkey, depending on the irrigation method, the weight of melon fruit ranged from 0.8 kg to 1.2 kg [38].

The nutritive values of melons are presented in the Table 5. The content of dry matter, total sugars, monosaccharides, ascorbic acid, total carotenoids, and polyphenols depended on both studied factors: drip irrigation and the cultivar. Drip irrigation combined with nitrogen fertigation significantly increased the concentration of studied components in relation to the control: for dry matter, by 1.7 points on average, total sugars by 14.1 points, monosaccharides by 17.3 points, ascorbic acid by 10.4 points, total carotenoids by 4.5 points, and total polyphenols by 10.8 points. In the study published by Ouzounidou et al. [45], melon fruit concentrated up to 5.1 g 100 g−<sup>1</sup> fresh weight of monosaccharides and from 0.8 g 100 g−<sup>1</sup> fresh weight to 4.0 g 100 g−<sup>1</sup> fresh weight of saccharose (total sugars). In the present experiment, similar results were obtained if we converted our figures into g 100 g−<sup>1</sup> fresh weight; the levels of total carbohydrates and monosaccharides in the edible parts of melon were significantly affected by the cultivar and irrigation treatments. Seledyn contained significantly more dry matter, total sugars, and monosaccharides than Melba. In

the studies presented by Wichrowska et al. [46], irrigation also had a positive effect on the concentration of reducing sugars and vitamin C in Cucurbitaceae, as in the present study. Melba contained significantly more ascorbic acid, total carotenoids, and polyphenols than Seledyn. Ouzounidou et al. [45] reported that the L-ascorbic acid concentration of melon fruit ranged from 13 mg 100 g−<sup>1</sup> fresh weight to 28 mg 100 g−<sup>1</sup> fresh weight. Substantially lower concentrations of this acid, in the range of 8 mg 100 g−<sup>1</sup> fresh weight to 13 mg 100 g−<sup>1</sup> fresh weight, were noted by Lin et al. [47]. The results of ascorbic acid concentration in the presented studies ranged between 14.5 mg 100 g−<sup>1</sup> fresh weight and 20.3 mg 100 g−<sup>1</sup> fresh weight and depended also on the irrigation treatment. Moreover, irrigation with fertilization increased the concentration of nutrients also in *Cucurbita maxima* Duch. [22].

**Table 4.** Influence of drip fertigation on the single fruit weight (kg) of Melba and Seledyn melon cultivars in the years 2013–2015.


<sup>1</sup> LSD = the lowest significant difference (Tukey's confidence half-interval) at *p* < 0.05; n.s.—not significant at *p* < 0.05.

**Table 5.** Influence of drip fertigation on the selected components of nutritive value of Melba and Seledyn melon cultivars (mean for the years 2013–2015).


<sup>1</sup> LSD = the lowest significant difference (Tukey's confidence half-interval) at *p* < 0.05; n.s.—not significant at *p* < 0.05.

#### **4. Conclusions**

The results of this study indicate that on loose sandy soil with low water capacity and nutrients, melon plants should be drip-fertigated with nitrogen, in order to obtain the best effects. As compared to the control (drip irrigation combined with nitrogen fertilization), the fertigation supplying nitrogen to the plants, used during the cultivation of two melon cultivars, Melba and Seledyn, on a loose sandy soil in Central Poland, significantly increased the total marketable fruit yield by 12.5% for Melba and by 13.6% for Seledyn. Compared to the control, drip fertigation also increased the weight of single fruits by 0.12 kg for Melba and by 0.19 kg for Seledyn. Drip irrigation combined with nitrogen fertigation significantly and positively influenced nutritive value, affecting the

increase in dry matter, total sugars, monosaccharides, ascorbic acid, total carotenoids, and polyphenols. Melba contained significantly more ascorbic acid, total carotenoids, and polyphenols than Seledyn, while Seledyn contained significantly more dry matter, total sugars, and monosaccharides.

Our studies fill a gap in the existing scientific literature and show that on a very light soil in a region with very low precipitation within Central Europe (Central Poland), the use of drip irrigation and nitrogen fertigation is effective for the cultivation of melon.

**Author Contributions:** Conceptualization, R.R., S.R. and A.F.; methodology, R.R., S.R. and A.F.; software, R.R., S.R. and A.F.; validation, R.R., P.-F.F. and P.S.; formal analysis, R.R., P.-F.F. and P.S.; investigation, R.R., S.R., A.F. and D.W.; resources, R.R. and S.R.; data curation, R.R., S.R., A.F. and D.W.; writing—original draft preparation, R.R., S.R., A.F., B.J., D.W., W.P., P.P., H.A.S., P.S. and D.L.; writing—review and editing, R.R., S.R., A.F., B.J., D.W., W.P., P.P. and P.S.; visualization, R.R., S.R., B.J., P.P., H.A.S., P.-F.F. and D.L.; supervision, R.R., S.R., B.J., P.P. and P.-F.F.; project administration, R.R.; funding acquisition, P.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Genotype** *×* **Environment Interaction of Yield and Grain Quality Traits of Maize Hybrids in Greece**

**Nikolaos Katsenios 1, Panagiotis Sparangis 1, Sofia Chanioti 2, Marianna Giannoglou 2, Dimitris Leonidakis 3, Miltiadis V. Christopoulos 2, George Katsaros <sup>2</sup> and Aspasia Efthimiadou 1,\***


**Abstract:** The interaction of genotype by the environment is very common in multi-environment trials of maize hybrids. This study evaluates the quantity and the quality of grain production and the stability of four maize genotypes in a field experiment that was conducted in five different locations for two years. In order to make a reliable evaluation of the performance of genotypes in the environments, principal components analysis (PCA) was used to investigate the correlation of the yield, soil properties and quality characteristics, while the additive main effects and multiplicative interaction (AMMI) analysis detected the narrow adaptations of genotypes at specific mega-environments. For the yield, AMMI analysis indicated that a group of five environments (ENV1, ENV8, ENV6 ENV10 and ENV9) gave higher yields than the mean value and at the same time had low first interaction principal components axis (IPC1) scores, indicating small interactions. Regarding protein and fiber contents, ENV1 and ENV2, gave the highest values and this could be attributed to the high concentration rates of nutrients like Mg, Ca and the soil texture (C). Specifically for the protein, the results of the analysis indicated that certain environment would provide more protein content, so in order to obtain higher grain protein, growers should grow in certain locations in order to improve the content of this quality characteristic, certain genotypes should be used in certain environments.

**Keywords:** genotype × environment interaction; maize; yield

#### **1. Introduction**

Maize (*Zea mays* L.) is a crop of major importance for the nutrition of the Earth's population. Thus, there has been an urgent need to increase its yield and its quality. There are two main factors that have approximately the same influence on yield increase; improved management practices along with plant breeding have made an impact on this cause. It is notable that their interaction is the one that made such a huge progress to this matter that neither could do alone to this extent [1]. Maize yields have been increasing over the years globally and according to recent data have been doubled from 1961 to 2002 [1–3]. Even before the use of hybrids, farmers used to breed plants that seemed to fit their needs, with good adaptation at their specific environmental conditions, while maintaining their quality and morphological traits. Wherever hybrids have been adopted there has been an increase in the maize yield. Even though there was a tendency to select the high yield hybrids, the need for overall stability and dependability favors the selection of hybrids with stress tolerance. The main focus of the new hybrids now is to aim for a high and stable yield in both favorable and unfavorable growing conditions [1]. Indeed, as Rosegrant et al. [4] mention maize production may suffer a huge input reduction since the available water

**Citation:** Katsenios, N.; Sparangis, P.; Chanioti, S.; Giannoglou, M.; Leonidakis, D.; Christopoulos, M.V.; Katsaros, G.; Efthimiadou, A. Genotype × Environment Interaction of Yield and Grain Quality Traits of Maize Hybrids in Greece. *Agronomy* **2021**, *11*, 357. https://doi.org/ 10.3390/agronomy11020357

Academic Editors: Nikolaos Monokrousos and Efimia M. Papatheodorou

Received: 23 January 2021 Accepted: 12 February 2021 Published: 17 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

resources will be limited especially in regions where irrigation is essential. So, it is implied that for the future of crop production it is a mandatory need, to include the environment as an important factor in the selection of better adaptive genotypes.

Climatic conditions along with soil characteristics are the main environmental factors that affect plant growth [5]. Atmosphere and soil water availability, soil temperature and composition can make an impact to the growth of the plants along with other factors such as soil pH and its nutrient status that can influence their development [6]. Soil plays a major role in the plant's life with emphasis on soil carbon, water quality and content. It is often related to problems that occur in plants such as nutrient deficiencies, water stress and toxicities. Its structure not only affects plant growth but also influences their ability to absorb nutrients and water [7]. Each environment has its own soil characteristics and climatic conditions that can affect the productivity of crop production. Thus, it is mandatory to take into consideration the effect of the environment while investigating the most suitable cultivation.

The genotype by environment (GE) interaction is a phenomenon recognized globally by everyone involved to the goal of crop improvement and maintenance. It refers to the various responses of genotypes across a wide range of environments [8,9]. Quantitative characteristics that are economically and agronomically important such as grain yield can be significantly affected by the GE interaction and can provide relief to the breeding actions that can be avoided, by reducing futile subsequent analyses, restricting the significance of questionable deductions and limiting the selection of superior genotypes [8,10–13]. In general, genotypes adapt differently to different environments and the evaluation of each one of them differs according to each purpose [14]. These evaluations in order to be valid are submitted to a series of multiple-environment trials (MET) in an advanced selection stage [11,15]. According to Lu'quez et al. [16], cultivars with high yield and better stability can be identified when growing cultivars in various environments. Every newly registered cultivar needs to be evaluated for adaptation for several years in many locations in order to be recommended to a specific area. To accomplish this procedure METs are conducted in many countries through varietal testing programs for more reliable results. The GE is the main interaction that needs to be evaluated. By submitting genotypes to a variety of environments, differential genotypic responses are recorded and can provide better identification of a superior and stable hybrid [17].

The environment, the genotype and the GE interaction are also responsible for variations in the quality properties of grains, including the color, the texture, the protein and the fiber content. Among the quality parameters, the protein content of the grains is highly affected by the environment [18]. The evaluation of different genotypes on quality traits associating with the improvement of the yield can also contribute to future breeding strategies.

The additive main effects and multiplicative interaction (AMMI) analysis is extensively used among agricultural researchers who want to evaluate the yield performance of different genotypes under multienvironment trials. This analysis is widely used because it contributes to better understanding of the complicated interactions between genotypes and environment, and it has high accuracy [19]. The AMMI analysis is a combination of ANOVA and PCA (principal component analysis) and a major output of the results is a biplot that presents the means of the genotypes used and their relation to the first PC [20]. This biplot is an effective tool to evaluate the GE interactions graphically [21]. The results of AMMI analysis are considered useful for the evaluation of yield stability of crops across different environmental conditions and for the determination of suitable environments for all examined genotypes [22–25].

The aim of this study was to evaluate four maize hybrids at five locations, for two years (the combination of year-location created the 10 environments), in order to investigate how maize yield and quality characteristics of the grain are affected by the GE interaction. Furthermore, principal components analysis (PCA) was used to investigate the correlation of the yield, soil properties and quality characteristics, while the AMMI analysis contributed to detect the narrow adaptations of genotypes at specific mega-environments.

#### **2. Materials and Methods**

#### *2.1. Experimental Site and Design*

The effect of GE interaction in relation to the final grain yield and the various quality characteristics were assessed. The field trials were conducted at 2019 and 2020 and a randomized complete block design was used with three replications. The experimental design had two main factors (genotypes and environments). The four maize genotypes used in the experimental fields were GEN1 (P0937, Pioneer Hi-Bred Hellas S.A.), GEN2 (DKC6050, K. and N. Efthymiadis S.A.), GEN3 (DKC6980, K. and N. Efthymiadis S.A.) and GEN4 (P2105, Pioneer Hi-Bred Hellas S.A.). P0937 hybrid belongs to the 500 FAO group with 120–125 days until maturity, it provides well-balanced plants, resistant to the northern leaf blight of maize, suitable especially for environments that traditionally produce good yield results. DKC6050 belongs to the 600 FAO group with 116–123 days until maturity. It produces average height plants with a very strong stem and root system and provides well-balanced plants. It has solid, filled until the top ears with 16-18 rows of kernels. It is considered to have great agronomic characteristics with high yield and good adaptation in many environments and with great quality grains suitable for human consumption. DKC6980 belongs to the 700 FAO group with 130–136 days until maturity. It produces high plants with a very strong stem and root system. It has big ears with 18-20 rows of kernels. It has stable yield even in high temperatures. P2105 is a hybrid that belongs to the 700 FAO group with 135–140 days until maturity, it provides well-balanced plants with a strong root system. It has a fast, lively growth and high dynamic production especially in early cultivated and well-drained fields. These four hybrids were selected because they are considered to have great characteristics in Greece's environmental variability.

The five locations were Giannouli, in the Prefecture of Larissa, Thessaly; Nea Tyroloi and Kamila, in the Prefecture of Serres, Central Macedonia; Kalamonas, in the Prefecture of Drama, Eastern Macedonia and Thrace and Koutso, in the Prefecture of Xanthi, Eastern Macedonia and Thrace, Greece. Each location–year combination created a different environment; thus 10 environments were used to evaluate the four genotypes.

The environments had different soil texture and variable microclimate conditions (Tables S1 and S2). Planting dates were between 22 and 25 March for the first year and 18 and 26 April for the second year and harvesting dates between 9 and 11 September for the first year and 19 and 21 September for the second year (Table S1). The plant density for all cultivars was 85,000 plants per hectare and the planting depth was 3.5 cm. The field plots (60 m2) consisted of 4 rows with spacing between rows 0.75 m. The measurement of yield was made at a length of 12 m in the 2 middle rows of each plot to avoid the border effect. All five locations are traditionally used for maize production in Greece since they appear to be suitable for this crop.

All field management procedures were standard to ensure the avoidance of deficiencies and the balancing of soil nutrients in all environments. After grain harvesting, the measurement of moisture content was conducted with a portable humidity meter in order to calculate the production per hectare adjusted to 15% humidity. Afterward, grain samples were directed for the determination and measurement of the quality characteristics.

#### *2.2. Soil Sample Analyses*

At the soil samples, the elements Ca, Mg, K and Na were determined by atomic absorption spectrometry after extraction using BaCl2 [26]. The measurements of Zn, Mn, Cu and Fe were conducted by atomic absorption spectrometry after extraction using DTPA [27]. The available B was determined using a spectrophotometer, using azomethine-H as the color (yellow) development reagent [28]. Total nitrogen was determined by the Kjeldahl method [29]. Organic carbon was determined with oxidization by K2Cr2O7 [30]. Available phosphorus was determined after extraction with NaHCO3 [31]. Cation exchange capacity was determined according to ISO 11260 [26]. Soil texture was determined using the method of Bouyoucos [32] and the soil taxonomy of USDA (1999). The moisture content

was determined in a furnace at 105 ◦C for 24 h. The value of pH was measured with a pH-meter equipped with a glass electrode in the saturated paste extract.

#### *2.3. Quality Characteristics of Harvested Corn Grains*

After their harvest, the corn grains were dried in the shade following the farming practices. The moisture content across all cultivars varied from 10.85 to 16.04% with an average of 12.74%.

The color of the corn grains was measured using Minolta Colorimeter (CR-300, Minolta Company, Chuo-Ku, Osaka, Japan). The lightness or brightness of the samples was indicated by the L value where 0–100 represents darkness to lightness color. The index a indicates the redness or greenness of the corn grains, with a positive a value representing more red color. The index b value represents the degree of the yellow-blue color, with a positive b value illustrating more yellow color.

The texture analysis was carried out by HD-Plus texture analyzer (Stable Micro Systems Ltd., Godalming, UK) and the Texture Expert Exceed Software for the data analysis. The determination of the textural characteristics of corn grains was performed by a puncture probe of 5 mm diameter. Probe speeds of 1 mm/s during the test, 2 mm/s for the pretest and 10 mm/s for the post-test were used throughout the study. All the measurements were performed at 25 ± 1 ◦C and the hardness of the corn seeds was determined and expressed at N.

The corn grains were grinded by using a grinding mill for the determination of the moisture, the ash, the total protein and the total crude fiber content. Ash and crude fiber content of corn flours were determined according to AOAC Official Method 923.03 and 984.04 (Weende Method), respectively and recorded manually. Total protein content analysis of corn flours was conducted by applying the Kjeldahl method (IDF 2008), using a Kjeldahl rapid distillation unit (Protein Nitrogen Distiller DNP-1500-MP, Raypa Spain).

#### *2.4. Statistical Analysis*

A two-way analysis of variance (ANOVA) was used to evaluate the effect of genotype, environments and their GE interaction on the cultivation and quality characteristics of maize hybrids. The experimental data were analyzed using IBM SPSS software ver. 24 (IBM Corp., Armonk, N.Y., USA). The comparisons of means were calculated using the Duncan test at the 5% level of significance (*p* < 0.05). Multivariate analysis was conducted by means of principal component analysis (PCA) by using STATISTICA 7 (Statsoft Inc., Tulsa, OK, USA). Additive main effects and multiplicative interaction (AMMI) analysis was conducted by using AMMISOFT version 1.0 (Soil and Crop Sciences, Cornell University, Ithaca, NY, USA).

#### **3. Results and Discussion**

#### *3.1. Grain Yield*

Yield (kg/ha) of maize was statistically significantly influenced by genotype, environment and their interaction (Table S3). Genotype and environment effects on maize yield are presented in Table 1. GEN1 had mean yield that was significantly different in all the tested environments. Its highest yield value was recorded in ENV10 and ENV9 (19,288 ± 289 and 19,087 ± 471 kg/ha respectively) and the lowest in ENV2 (12,805 ± 1361 kg/ha). GEN2 presented similar results; the highest mean yield was recorded in ENV10 and ENV9 (18,244 ± 182 and 18,113 ± 86 kg/ha respectively) but the lowest mean yield in this case was recorded in ENV4 and ENV5 (13,617 ± 370 and 13,657 ± 51 kg/ha respectively). On the other hand, GEN3 presented its highest mean yield in ENV10 where it recorded 20,032 ± 179 kg/ha, and significantly lower results in ENV4 (13,220 ± 569 kg/ha). GEN4 had high yields in ENV6 (20,070 ± 346 kg/ha) and the significant lower value was presented in ENV4 (12,077 ± 166 kg/ha). As for the environments, in ENV4, ENV8 and ENV9, GEN1 presented significant higher mean yields. ENV4 had also a statistically significant performance in GEN2 along with ENV2 and ENV3. Environments ENV1, ENV6 and ENV7 showed significant high performance with GEN4. All the environments presented great

results when GEN3 was used except ENV6 and ENV7, which had their lowest performance with this genotype.

**Table 1.** Effect of genotype and environments on yield and quality characteristics (protein, fiber, color parameters, texture and ash content) of corn grains.



**Table 1.** *Cont.*

Mean value of three replicates ± standard deviation. Values with different capital letter (A, B, C, D, E, F, G, H, I, J) denotes significant difference between environments, and small letter (a, b, c, d) denotes significant difference between genotypes in each environment according to the Duncan's multiple range test at *p* < 0.05. Where there are no letters, no significant differences were recorded.

#### *3.2. Quality Characteristics of Harvested Corn Grains*

The performance of genotypes on environments for quality characteristics (color parameters, texture, ash, protein and fiber content) of the harvested corn grains is presented in Table 1. The performance (color parameters) for corn grains varied across different environments. The lightness (L) across all cultivars ranged from 34.28 to 78.23, the yellow index b ranged from 34.47 to 58.90 and the red index a ranged from 3.61 to 8.80. For L, a and b color parameters, there was a significant difference between genotype (*p* < 0.001), environments (*p* < 0.001) and for their GE interaction, except for the interaction on the red index. The highest values of L color parameter of corn grains were obtained from the environments

ENV2, ENV4, ENV6, ENV8 and ENV10 and the lowest from the environments ENV1, ENV3, ENV5 ENV7 and ENV9. According to the average of the tested environments, the highest values of L, a and b color parameters were achieved from genotype GEN2, and the lowest corresponding values were obtained from the genotypes GEN1 and GEN4 (Table 1).

The texture of the harvested corn grains was influenced by the environmental and genotype effects (*p* < 0.001) (Table 1). Texture (hardness) of corn grains varied between 12.53 and 26.28 N across the cultivars and environments. The hardness of corn grains at ENV1 (21.38 N), ENV4 (18.75 N), ENV2 (18.64 N), ENV6 (18.63 N), ENV5 (18.33 N) and ENV10 (18.10 N) was relatively higher than the one at ENV3 (13.47 N) and ENV8 (13.38 N). According to the average of the tested environments, the highest values of hardness were achieved from genotypes GEN2 (20.72 N) and GEN1 (20.23 N), and the lowest corresponding values were obtained from the genotypes GEN3 (18.50 N) and GEN4 (17.20 N) (Table 1).

The ash content of corn grains was significantly affected by genotype (*p* < 0.001), environments (*p* < 0.001) and their GE interaction (*p* < 0.001) (Table 1). The ash content across all cultivars ranged from 0.73 to 2.40%. The highest ash content was found in corn grains obtained at the environments of ENV1 (1.51%) and ENV4 (1.50%) and the lowest at ENV8 (1.07%). The ash content of corn grains from GEN2 (1.39%) and GEN4 (1.38%) was significantly higher than the one from GEN1 (1.30%) and GEN3 (1.23%).

The genotype (*p* < 0.001), the environments (*p* < 0.001) and their GE interaction (*p* < 0.001) highly influenced the protein content of corn grains (Table 1). The protein content across all genotypes varied from 6.13 to 8.96%. The highest protein content was found in corn grains obtained at the ENV1 (8.63%). The protein content of corn grains from GEN3 (7.59%) and GEN4 (7.45%) was significantly higher than the ones from the other genotypes. Protein content is a primary quality indicator for corn grains. Mut et al. [33] and Peterson et al. [34] reported that the grain protein content changed from 3 to 4% and 10.0 to 18.0%, respectively within different oat genotypes cultivated in different environments. The protein content was mainly affected by the environment rather than the genotype. This finding is in accordance with other scientific studies [18,35,36]. The protein content of corn grain is illustrated the quality of corn flour and is a desirable trait for the food industry.

The fiber content of corn grains was significantly influenced by genotype (*p* < 0.01), environments (*p* < 0.001) and their GE interaction (*p* < 0.001) (Table 1). The fiber content across all cultivars ranged from 1.37 to 4.08%. The highest fiber content was found in corn grains obtained at the environments of ENV2 (3.65%) and ENV1 (3.54%) while the lowest at ENV7 (2.01%). The fiber content of corn grains from GEN2 (2.97%) and GEN1 (2.85%) was significantly higher than others. It was observed that the effect of environment on the fiber content of corn grains was stronger as compared to the genotype. This finding is in accordance with other scientific studies [18,37].

#### *3.3. Correlation and Evaluation of the Yield, Soil Properties and Quality Characteristics vs. the Genotype and Environment on Maize Cultivation*

To investigate the correlation of the yield, soil properties and quality characteristics by using four different genotypes of maize hybrids at ten different environmental conditions, principal components analysis (PCA) was used (Figure 1). Each point on the loading plot represented the contribution of a variable (yield, soil properties: clay, silt, sand, pH, organic matter, total nitrogen, CaCO3, K, Ca, Mg, P, Fe, Cu, Zn, Mn, B and quality characteristics: color, texture, moisture, ash, protein and fiber content) to the score, while each point on the score plot represented a tested sample. The first principal component (PC1) described 41.38% of the variation of extraction experiments, whereas the second principal component (PC2) 25.63% respectively, so that they contributed 67.01% of the total variation of extraction experiments.

**Figure 1.** Biplot of principal component analysis of the four different genotypes of maize hybrids at five different environmental conditions. Code of different environments of maize hybrids on different genotypes used on principal components analysis (PCA) is listed as follows: Environments: (1) ENV1, (2) ENV2, (3) ENV3, (4) ENV4, (5) ENV5, (6) ENV6, (7) ENV7 (8) ENV8, (9) ENV9 and (10) ENV10 and Genotypes: (a) GEN1, (b) GEN2, (c) GEN3 and (d) GEN4.

According to the PCA plot, the texture, fiber, protein, clay, Mg, Ca, pH and OM had a negative effect on PC1 and the total nitrogen, silt, CaCO3, Fe and K had a negative effect on PC2, while the sand, P and Zn had a positive effect on PC1 and the Mn, silt and yield had a positive effect on PC2. Furthermore, there are correlations between the Mg and fiber, clay and protein, pH and texture, and between L, B, K, Cu and CaCO3. Based on PCA score plot of the tested samples, four main groups of samples were noted. The groups are (a) 1a, 1b, 1c, 1d, 2a, 2b, 2c, 2d (b) 5a, 5b, 5c, 5d, 6a, 6b, 6c, 6d (c) 10a, 10b, 10c, 10d (d) 3a, 3b, 3c, 3d, 8a, 8b, 8c, 8d and (e) 7a, 7b, 7c, 7d, 9a, 9b, 9c, 9d.

All five testing locations are suitable for maize production in Greece. However, they have different soil conditions and rainfall, influencing the yield and the quality characteristics of corn grains. The samples of group (a) confirmed that ENV1 and ENV2 were the most effective environments for all the tested hybrids, giving corn grains with the highest protein and fiber content. These findings could be attributed to the enhanced soil fertility of ENV1 and ENV2 having the highest concentration of nutrients including Mg, Ca, clay and silt. Many studies demonstrated that the protein content was mostly affected by the environment, indicating its sensitivity to the environment [18,35,36,38] and that the soil nutrient supply affected positively the yield and the quality characteristics of the crop products [39].

The samples of group (b) indicated that ENV5 and ENV6 showed good soil conditions in terms of nutrients resulting in high yields. This finding is in accordance with other studies indicating that any higher nutrient uptake by the plant can result in higher yields [18,40]. The samples of group (c) showed that ENV10 resulted in grains with the maximum hardness concerning their texture essential soil indices such as pH and OM compared with the other environments. Moreover, the samples of group (d) showed that ENV3 and ENV8 had similar characteristics in terms of protein and fiber content growing on a

Fe-, P- and N- and sand-rich environment. The samples of group (e) indicated that ENV7 and ENV9 had similar characteristics in terms of yield and quality properties growing on a Zn-rich environment. Concluding, the results depicted by principal components analysis are in agreement with those discussed above.

#### *3.4. GE Interaction for Yield, Protein and Fiber Content*

According to the ANOVA, genotypes (GEN), environments (ENV) and their interaction (G×E) gave statistically significant differences (*p* < 0.001) concerning the yield measurement. Moreover, the highest percentage of variation explained by ENV (68.89%), followed by the G×E (27.15%) effect, while GEN explained (3.95%) of the variation (Table S3).

Figure 2 shows that GEN3 had the highest mean yield, followed by GEN1, GEN4 and GEN2. Among these, GEN4 was the hybrid with the lowest score of the first interaction principal components axis (IPC1). The great score values of IPC1 mean that these genotypes are adapted to certain environments [24]. As for the environments, ENV10 presented the highest mean yield (18,862 kg/ha) with an IPC1 score close to zero, indicating small interactions and ENV4 the lowest yield (13,227 kg/ha). ENV1, ENV6, ENV8 and ENV9 had IPC1 values close to zero and yield higher than the mean value (17,941, 18,839, 17,691 and 18,446 kg/ha respectively).

**Figure 2.** Additive main effects and multiplicative interaction (AMMI) biplot presenting mean grain yield (kg/ha) and the first interaction principal components axis (IPC1) of 4 genotypes (red) evaluated in 10 environments (blue).

Based on AMMI1 model, GEN3 and GEN1 resulted in the highest narrow adaptations and these were the best adapted genotypes of the two mega-environments delineated (Table 2). The first one consisted of ENV5, ENV2, ENV6, ENV1, ENV10 and ENV3 in which GEN3 was the best adapted genotype. The other one consisted of ENV8, ENV9, ENV4 and ENV7 with GEN1 presenting better results. Genotypes and environments had been listed according to their IPC1 order (Table 3), resulting that top and bottom performances have the opposite GE pattern [33]. For instance, GEN3 had positive GE with the environments such as ENV5 and ENV2 and a negative GE with environments like ENV7 and ENV4.


**Table 2.** AMMI family models for the grain yield, protein content and fiber content dataset, winning genotypes and the numbers of mega-environments.

**Table 3.** Ranking of the genotypes and the environments based on their IPC1 scores for grain yield, protein content and fiber content.


The first two principal components in AMMI analysis were significant (*p* < 0.001), explaining 82.13% of GE (44.49% IPC1 and 37.64% IPC2) of interaction variation (Table S3). According to the biplot of the first (IPC1) and the second (IPC2) interaction principal components (Figure 3), GEN2 had a positive interaction with five out of ten environments (ENV3, ENV4, ENV8, ENV9 and ENV10). Generally, all genotypes were located far from the biplot origin and contribute to the G×E interaction for yield. AMMI analysis is widely used

for the evaluation of maize hybrids yield in multienvironment field trials [20,24,25,41–43], and grain yield of wheat varieties [22,36,44–46], seed yield of oilseed rape [23], nutritional composition of sweet potato [38], yield of sugarcane [47] and yield of chickpea [48].

As for the protein content, according to the ANOVA, genotypes (G), environments (E) and their interaction (GxE) gave statistically significant values (*p* < 0.001). Moreover, the highest percentage of variation explained by ENV (66.90%) and G×E (29.32%) effects, while G explained the rest of variation (3.79%) (Table S3). Figure 4 shows that all genotypes had protein content percentages close to the mean value, with IPC1 values far from zero. ENV1 was the environment that presented the best results for the protein content (8.63%) of corn grains, while ENV9 presented the lowest one (6.54%). ENV10, ENV8 and ENV9 were the environments that presented the most stable results in terms of protein content.

According to AMMI1 model, GEN3 and GEN2 resulted in the highest narrow adaptations delineating two mega-environments (Table 2). The first one consisted of ENV7, ENV3, ENV9, ENV8 and ENV1, in which GEN3 was the best adapted genotype and the other one consisted of ENV10, ENV5, ENV4, ENV2 and ENV6, where GEN2 was the better suited genotype.

The first two principal components in AMMI analysis were significant (*p* < 0.001), explaining 83.93% of GE (48.61% IPC1 and 35.32% IPC2) of interaction variation (Table S3). According to the biplot of the first (IPC1) and the second (IPC2) interaction principal components (Figure 5), GEN1 had a large positive interaction with ENV4 and ENV2, GEN2 had a large positive interaction with ENV6, while GEN4 and GEN3 had a positive interaction with ENV8 and ENV9 and ENV3 respectively. Likewise the yield, all genotypes are located far from the biplot origin and contribute to the G×E interaction for yield.

**Figure 3.** AMMI biplot presenting the second interaction principal components axis (IPC2) against the first interaction principal components axis (IPC1) scores for grain yield (kg/ha) of 4 genotypes (red) evaluated in 10 environments (blue).

**Figure 4.** AMMI biplot presenting mean protein content and the first interaction principal components axis (IPC1) of 4 genotypes (red) evaluated in 10 environments (blue).

**Figure 5.** AMMI biplot presenting the second interaction principal components axis (IPC2) against the first interaction principal components axis (IPC1) scores for protein content of 4 genotypes (red) evaluated in 10 environments (blue).

Regarding fiber content, according to the ANOVA, genotypes (G), environments (E) and their interaction (GxE) gave statistically significant values (*p* < 0.001). Moreover, the highest percentage of variation explained by ENV (56.43%) and G×E (39.15%) effects, while G explained the rest of variation (4.42%) (Table S3). Figure 6 presents that GEN2 had the highest value, followed by GEN1, GEN3 and GEN4. GEN1 had an IPC1 score close to zero, indicating small interactions. ENV6 scored an IPC1 value near 0, and at the same time had fiber content slightly higher than the mean value. ENV2 was the environment that presented the best results for the fiber content (3.65%) of corn grains, while ENV7 presented the lowest one (2.01%).

According to AMMI1 model, GEN2 and GEN4 resulted in the highest narrow adaptations defining two mega-environments (Table 2). The first one consisted of ENV10, ENV7, ENV5, ENV2, ENV6, ENV8, ENV9 and ENV4 in which GEN2 was the genotype that presented better results and the other one consisted of ENV1 and ENV3 that had GEN4 as the better suited genotype. The AMMI statistical model has been used to evaluate the effects of genotypes, environments and their interaction for quality characteristics, like iron and zinc concentrations in the grain of maize [49], protein and tryptophan in maize [50], nutritional composition (protein, β-carotene, iron, zinc, starch and sucrose) in sweet potato [38] and vitreousness, SDS sedimentation test, yellow pigment index, protein content and test weight in durum wheat [46].

The first two principal components in AMMI analysis were significant (*p* < 0.001), explaining 90.79% of GxE (55.20% IPC1 and 35.59% IPC2) of interaction variation (Table S3). According to the biplot of the first (IPC1) and the second (IPC2) interaction principal components (Figure 7), GEN2 had a large positive interaction with ENV7, GEN4 had a large positive interaction with ENV3, while GEN3 had a positive interaction with ENV10.

**Figure 6.** AMMI biplot presenting mean fiber content and the first interaction principal components axis (IPC1) of 4 genotypes (red) evaluated in 10 environments (blue).

**Figure 7.** AMMI biplot presenting the second interaction principal components axis (IPC2) against the first interaction principal components axis (IPC1) scores for fiber content of 4 genotypes (red) evaluated in 10 environments (blue).

#### **4. Conclusions**

The evaluation of yield results of different genotypes, under different environmental conditions is a complicated issue, as a lot of parameters have to be considered, in order to lead to reliable results. In such experiments, often the AMMI analysis was used, which provides valuable information that contributes to the understanding of the G×E interaction. In this study, the GEN effect explained a low percentage of the variation and could not lead to the selection of a stable genotype for all environments. However, the results of the AMMI analysis contributed to dividing the region into mega-environments and introduce the most suitable genotype for every environment.

Concerning the yield, based on the AMMI1 model, GEN3 and GEN1 resulted in the highest narrow adaptations and these were the best adapted genotypes of the two mega-environments delineated (the first consisted of ENV5, ENV2, ENV6, ENV1, ENV10 and ENV3 and the second consisted of ENV8, ENV9, ENV4 and ENV7, respectively). A group of five environments (ENV1, ENV8, ENV6 ENV10 and ENV9) gave higher yields than the mean value and at the same time had low IPC1 scores, which indicated that they gave high yield with all the genotypes used. Regarding the grain quality, GEN3 and GEN2 for the protein content and GEN2 and GEN4 for the fiber content resulted in the highest narrow adaptations delineating two mega-environments. Specifically for the protein, the results of the analysis indicated that in order to obtain higher protein content, certain genotypes should be used in certain environments. It is important to note that ENV1 and ENV2 (location Giannouli for years 2019 and 2020 respectively, gave the highest values concerning protein and fiber content. These findings could be attributed to the high concentration rates of nutrients like Mg, Ca and the soil texture (C).

The results of this study suggest that the target to increase the quantity and quality of grain yield of maize hybrids is a very challenging issue, due to the high G×E interaction, which can be implemented by exploiting positive GE interactions, by dividing the environment into mega-environments.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2073-439 5/11/2/357/s1, Table S1. Coordinates, altitude, soil texture and cultivation information for the ten environments. Table S2. Climatic conditions of the 10 examined environments during the cultivation period (March-September). Table S3. AMMI analysis of variance for grain yield, protein content and fiber content of 4 genotypes evaluated in 10 environments.

**Author Contributions:** Conceptualization, N.K. and A.E.; Methodology, N.K., P.S., S.C., M.G., D.L., M.V.C., G.K. and A.E.; Software, N.K., P.S., S.C. and M.G.; Validation, D.L., G.K. and A.E.; Formal Analysis, N.K., P.S. and S.C.; Investigation, N.K., S.C., M.G., D.L. and M.V.C.; Resources, N.K., P.S., S.C., M.G., D.L. and M.V.C.; Writing—Original Draft Preparation, N.K., P.S., S.C. and M.G.; Writing— Review and Editing, N.K., P.S., S.C., M.G., D.L., M.V.C., G.K. and A.E.; Supervision, A.E.; Project Administration, N.K., D.L. and G.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interests.

#### **References**

