SOIL ATTRIBUTES AND WHEAT PERFORMANCE IN A GEOREFERENCED SEED PRODUCTION FIELD

The aim of this research was to study the linear correlations and spatial distribution between soil attributes and productivity and physiological quality of wheat seeds. Soil samples were collected at georeferenced points in an area of 41 ha with a rectangular grid of 100 × 100 m to determine soil fertility. Grain productivity, number of ears m, protein content, and physiological seed quality were assessed. The results were submitted to Pearson’s correlation analysis, descriptive statistics, and geostatistics. Phosphorus content presented a high spatial variability, with a coefficient of variation of 97.79%. Protein content, number of ears, and SMP index presented low values of coefficient of variation, indicating their low spatial variability. Physiological seed quality was not correlated with soil attributes. Wheat productivity showed a strong correlation between calcium and soil organic matter contents, as verified by analyzing the spatial distribution maps. The analysis of the spatial distribution maps of soil nutrients is a tool that can contribute to farmers’ decision making in the management of crop fertilization aiming at increasing productivity.


INTRODUCTION
Precision agriculture is a set of tools to help farmers to manage field processes using a series of information about the area, such as digital models and satellite images for soil fertility and productivity. This tool aims at managing soil, plant, and environmental spatial and temporal variability to achieve a high productivity and improved input-use efficiency, reducing production costs (Luz et al., 2013).
The collection of georeferenced soil samples allows spatializing soil chemical attributes with geostatistics tools, frequently used to identify zones with chemical restrictions that may be limiting the yield. However, the characterization and spatialization of soil physical attributes in commercial areas have been little used due to the inherent difficulties for collecting and analyzing a large number of samples (Amado et al., 2009).
Soil compaction reduces root growth due to an increase in soil penetration resistance, which can lead to crop productivity losses related to the restriction of water and nutrient to plant roots (Bergamin et al., 2010).
Soil chemical attributes have a great spatial and temporal variability, which can be attributed to intrinsic and extrinsic factors, i.e. pedogenetic factors of soil and factors related to its management. This variability in the soil may have a high correlation with physiological seed quality since nutrients play specific functions in plants and in their metabolic processes, which can be affected by nutrient content and its availability in the soil. However, working with the correlation between soil nutrient contents and physiological quality of soybean seeds, Mondo et al. (2012) did not find a significant correlation between the contents of phosphorus, potassium, calcium, magnesium, and pH with germination, first germination count, and accelerated aging.
The study of the correlation between soil attributes and plants is important to establish and quantify soil attributes that have a greater influence on crop productivity, as mentioned in several studies (Amado et al., 2009;Rodrigues et al., 2012;Santi et al., 2012;Mattioni et al., 2013;Gazolla-Neto et al., 2016). This study fits as a way to manage a field of seed production since it is a crop with a high added value and needs a special attention to obtain seeds with a high physiological quality.
In this context, the aim of this research was to study the linear correlations and spatial distribution between soil attributes and productivity and physiological quality of wheat seeds by using precision agriculture techniques.

MATERIAL AND METHODS
This study was carried out in an area of 41 ha of wheat seed production of the company Granello Sementes in Lavras do Sul, RS, Brazil (30°43′18.6″ S and 53°56′44.4″ W), in the 2013/2013 season. The relief of the area is little wavy, with soil classified as a Neossolo Litólico eutrófico (EMBRAPA, 2013) (Lithic Quartzipsamment, USDA Soil Taxonomy). According to Köppen classification, regional climate is type Cfa.
From the outline map of the area, a sampling grid was elaborated with sampling points spatialized in 100 × 100 meters, characterizing one point per hectare and totaling 41 collection points. For determining soil and seed attributes, four subsamples were collected, being one subsample at the georeferenced point (central sampling point) and the others collected at points equidistant around the point in a 15-m radius (Figure 1). The central sampling points were defined based on the systematic sampling method, following a regular sampling grid, where points are marked along a predefined course and based on the field outlines (Mattioni et al., 2011). Samples were collected in the interrow, at a depth of 0.00-0.10 m, as indicated by CQFS (2016).
This study was carried out in a consolidated notillage area, in which the production system is characterized by the succession of the following crops: Sowing was carried out on June 14, 2013, to reach a population of 300 plants m −2 . In the base fertilization, 280 kg ha −1 of the NPK formula 5-20-20 was used according to soil analysis. Topdressing nitrogen fertilization was fractionated into two applications, one at the tillering stage (90 kg ha −1 urea) and other at stem elongation stage (190 kg ha −1 urea). A fertilization with potassium chloride (100 kg ha −1 ) was also performed at the stem elongation stage. The phenological scale is in accordance with Rodrigues et al. (2011).
The determination of soil mechanical penetration resistance was carried out by means of a Falker PLG 1020 electronic penetrometer. The measurement was performed at each central point of the sampling grid and at the three lateral points. The layer of 0.00 to 0.40 m was analyzed on November 04, 2013, and the results were presented in layers of 0.00-0.10, 0.11-0.20, 0.21-0.30, and 0.31-0.40 m.
The assessment of soil chemical attributes (calcium, magnesium, phosphorus, potassium, organic matter (OM), cation exchange capacity (CEC), SMP index, aluminum, sulfur, sand, and clay) was performed after wheat harvesting according to the indications of the Comissão de Química e Fertilidade do Solo (CQFS, 2016).
In order to determine the productivity and physiological seed quality, plants were collected in four 1m 2 subsamples at each point of the sampling grid, as described previously, forming an average sample of the point.
Physiological seed quality was assessed by the following tests: Germination (G): conducted with eight replications of 50 seeds. Sowing was conducted on paper substrates previously moistened with distilled water in the proportion of 2.5 times the weight of the dry paper and maintained in a germinator at 20 °C. The assessments were carried out at four and eight days after sowing, according to the Rules for Seed Testing (Brasil, 2009) and the results were expressed as a percentage of normal seedlings.
First germination count (FGC): conducted together with the germination test and assessment at four days after sowing.
Electrical conductivity (EC): performed by the mass system according to the methodology proposed by the vigor committee of the Association of Official Seed Analysts, described by Marcos Filho et al. (1987). Four samples of 50 seeds, previously weighed in a 0.0001 g precision scale, were placed in plastic cups containing 50 mL deionized and distilled water. The cups were maintained in an environment at 25 °C and readings were performed after soaking for 24 hours. The obtained average values were expressed in µS cm −1 g of seeds −1 . The test was performed on February 25 and 26, 2014.
Engenharia Agrícola, Jaboticabal, v.38, n.4, p.624-633, jul./ago. 2018 Accelerated aging (AA): performed in Gerbox boxes with a metallic screen. A volume of 40 mL distilled water was added to the bottom of each box and seeds were evenly distributed on the screen in a single layer. Subsequently, these boxes were capped and conditioned in a BOD incubator at 41 °C for 72 hours. After this period, seeds were submitted to the germination test (Delouche & Baskin, 1973).
Field seedling emergence (SE): carried out with four subsamples of 50 seeds for each sampling point, distributed in a 1.0 m long furrow and a depth of 2.5 cm. Sowing was carried out in soil beds at the time recommended for cultivation. The emerged seedling count was performed at 14 days after sowing.
Number of ears: obtained by counting the ears in the four subsamples of one square meter and the result was expressed in ears m −2 .
Productivity: the ears in the four subsamples of one square meter at each sampling point were harvested and threshed, moisture corrected to 13%, and the result expressed in kg ha −1 .
Pearson's linear correlation coefficients were determined using the WINSTAT statistical program (Machado & Conceição, 2007). Spatial variability was analyzed using the classification criterion for coefficient of variation (CV) proposed by Wilding & Dress (1983), which defines as low the variability of attributes with a CV value lower than 15%, moderate the variability with a CV between 15 and 35%, and high the variability with a CV value higher than 35%.
Digital models were elaborated by means of the software Sistema Agropecuário CR -Campeiro 7 (Giotto et al., 2004). Kriging was the geostatistical method of interpolation used for elaborating the digital models. Parameter estimation was carried out by the maximum likelihood method.
Pearson's linear correlation of soil and wheat attributes is represented by values ranging from −1 to +1, so that according to the criterion proposed by Dancey & Reidy (2013), positive or negative values below of 0.3 indicate a low correlation, between 0.3 and 0.7 indicate a moderate correlation, and above 0.7 indicate a strong correlation.
The degree of spatial dependence (DSD) of semivariograms was assessed considering the intervals proposed by Cambardella et al. (1994): DSD < 25%strong spatial dependence; 25% < DSD < 75%moderate spatial dependence; and DSD > 75%weak spatial dependence. Table 1 shows the descriptive statistics of soil attributes. Phosphorus was the soil attribute that presented the highest spatial variability, with a CV of 97.79%, according to the spatial variability criterion proposed by Wilding & Dress (1983). Alves et al. (2014) and Gazolla-Neto et al. (2016) found similar results for phosphorus, which presented coefficients of variation of 95.35 and 102%, respectively.

RESULTS AND DISCUSSION
This high spatial variability of phosphorus could be explained by its low mobility in the soil. Thus, its superficial accumulation occurs in no-tillage areas, caused by the non-disturbance of soil and fertilization of the row, hindering its loss by erosion.
Calcium, magnesium, and organic matter presented a moderate spatial variability (Table 1). Mondo et al. (2012) found similar results, with organic matter and magnesium presenting a moderate variability, but with a low variability of calcium. Table 1 shows a strong degree of spatial dependence for Ca, OM, S, sand and clay contents, CEC, and SMP index. According to Cambardella et al. (1994), attributes that have a strong spatial dependence are more influenced by the intrinsic soil properties such as texture and mineralogy, while those with a week spatial dependence are more influenced by external factors such as fertilizer application, soil tillage, and cultivation, i.e. soil management. The other variables presented a moderate spatial dependence, corroborating the results found by Gazolla-Neto et al. (2016).
The range is the main parameter provided by geostatistics, which is the distance at which a variable signals spatial continuity and, from this distance, the spatial behavior of the variable becomes totally random (Lemos Filho et al., 2008). On the other hand, determinations performed at lower distances are correlated to each other, allowing interpolations for smaller spacing than those sampled (Andriotti, 2013). According to the data analysis observed in Tables 1 and 2, range values of the spatial dependence varied from 200 to 600 m. Therefore, the sampling grid was adequate since it was smaller than 200 m. In general, soil attributes have larger ranges, while plant attributes present smaller ranges.   Regarding the crop attributes (Table 2), the descriptive statistics showed for productivity a minimum value of 2739 kg ha −1 , maximum of 5378 kg ha −1 , and an amplitude of 2639 kg ha −1 , i.e. a potential to produce 5378 kg ha −1 , but with an average of 4243 kg ha −1 . Therefore, it is necessary to observe and correct the factors that most influence wheat productivity. Productivity presented a CV of 12.44%, which shows a low spatial variability in the study area. The number of ears presented minimum, maximum, and average values of 333.0, 479.75, and 402.571 ears m −2 , respectively, with a CV of 10.11%, which means a low spatial variability (Table 2).
In addition, protein content, number of ears m −2 , and SMP index presented low CV values, indicating a low spatial variability.
The attributes of physiological seed quality presented a low spatial variability ( Table 2). The average germination was 95%, with values ranging from 89 to 98% and a CV of 1.75%, which is considered as low. Germination, first germination count, and field seedling emergence presented an amplitude between the minimum and maximum values of 9, 29, and 37%, respectively. The different treatments during cultivation may contribute to the production of seeds with a high physiological quality.  Soil attributes that best correlated with wheat productivity were calcium, magnesium, organic matter, and CEC. Rodrigues et al. (2012)  and nutrients in points of higher depths due to its superficial rooting (Miguel et al., 2010).
Wheat productivity showed a strong correlation between calcium and soil organic matter, as shown in the spatial distribution maps (Figure 2). This correlation between productivity and organic matter shows its important role in improving soil structure and porosity, contributing to an increased productivity. Moreover, productivity presented a negative correlation with soil penetration resistance in the layers of 0.0-0.10 and 0.11-0.20 m, i.e. the higher the soil penetration resistance is, the lower the productivity. In fact, an increase in soil compaction reduces root growth due to an increase in soil penetration resistance, which can lead to crop productivity losses due to a restriction of water and nutrient to plant roots (Bergamin et al., 2010).  (Table 3).
Engenharia Agrícola, Jaboticabal, v.38, n.4, p., jul./ago. 2018 Regarding wheat technological quality, sulfur is essential for chlorophyll formation, being a constituent of gluten-forming proteins, in addition to stabilizing polymers of glutenin by sulfur bridges, which determine mass elasticity (Rodrigues & Teixeira, 2010). This information is in accordance with the spatial distribution trend observed in sulfur and protein maps, which have the same spatial distribution trend (Figure 2).
The bottom of the maps of digital models of spatial distribution presented higher contents of Ca, Mg, OM, S, and P, corresponding to a higher wheat productivity and number of ears m −2 (Figure 2). Productivity map is an important tool in decision-making and analysis of agricultural performance at a property level (Amado et al., 2009). In this sense, productivity had a significant and positive correlation with OM content since it is related to a higher soil water retention, increasing its availability to crops as the water deficit affects metabolism and decreases plant growth (Mondo et al., 2012).
Pearson's correlation analysis presented a positive correlation between productivity and clay content (Table   3) and a negative correlation between productivity and sand content (Figure 2). In the region with the highest clay content is the highest contents of Ca and OM, leading to a higher productivity in this area since cation retention depends on clay content.
The significant correlation found between productivity and calcium and magnesium may be an important tool in wheat production since these subareas can be treated separately according to the most influential elements of productivity at low or high contents. The Spatial distribution maps showed that soil penetration resistance (Figure 3) might have influenced wheat seed productivity since in the region with the highest compaction in the three soil layers, a lower productivity was observed. Girardello et al. (2014) observed similar results in a study that aimed to assess the efficiency of chisel plows and their effect on soybean productivity. The lowest contents of phosphorus and calcium were observed in the region with the highest soil compaction (Figure 3), leading to a lower wheat productivity, i.e. soil compaction affected nutrient absorption and limited infiltration and redistribution of water. Similarly, Amado et al. (2009) found that the zone of lower productivity presented characteristics of higher penetration resistance, i.e. a higher soil compaction.

CONCLUSIONS
The wheat cultivation area presents spatial variability of physiological seed quality although no correlation was observed with soil attributes.
Wheat productivity is strongly correlated with calcium and soil organic matter contents.
The analysis of spatial distribution maps of soil nutrients is a tool that can contribute to farmers' decision making in the management of crop fertilization aiming at improving seed productivity.