Quality of Soybean Seeds Cultivated on Different Potassium Fertilization Management

Soybean is one of the crops worldwide cultivated, and although it is usually commercialized quantitatively, qualitative characteristics of its production are highlighted, particularly oil and protein content, which is important for human and animal nutrition besides higher industrial yields in the synthesis of its derivatives. This study assessed the quality seed changes of soybean cultivated under different potassium rates in an Oxisol under no-tillage system in Floresta, Paraná State, Brazil. The experiment designed was in complete randomized blocks composed of a cross factorial (5 × 2) with four replicates. It was carried out in two growing seasons (2016/2017 and 2017/2018) totaling 40 experimental units. The rates (0, 40, 80, 120, and 160 kg ha of K) corresponded to the first factor and sowing fertilization (0 and 30 kg ha of K) was the second factor. Seed electrical conductivity, water content, seed K leaching, seed K content, oil and protein content, seed density, seed mass and yield were measured. The results indicated that K application for soybean may promote better quality seeds production, since electric conductivity, oil and water content and yield have increased in some conditions, although the sowing fertilization did not influence.


Introduction
Soybean [Glycine max (L).merr] is the most important cash crop in Brazil and in several countries around the world.In Brazil, soybean has been cultivated in all regions and its production complex is the main agribusiness productive chain.Brazilian soybean production is the second world's largest (CONAB, 2017).Due to versatility of its derivatives the demand for soybean grain is increasing every year by the external and domestic markets, consequently there is an expansion of the area cultivated with soybean on new agricultural frontiers.In addition, many investments are made to supply technologies to increase production by area (Linzmeyer Junior et al., 2008).
According to Foloni and Rosolem (2008), soybean usually responds to potassium fertilization in tropical soils.On the other hand, some authors have shown that there is lack or no response in conditions above the critical level of K in soil (Guareschi et al., 2008;Bernardi et al., 2009;Gonçalves Júnior et al., 2010).Under these conditions, responses are subtle and often do not significantly increase yield.Some authors have found significant differences for K application in the soil by analyzing soybean seeds as in germination tests (Toledo et al., 2011), on K contents exported by seed (Serafim et al., 2012), on protein content and seed mass (Khan et al., 2004) and vigor seed (Petter et al., 2014).Nevertheless, the increase in K seed exported would be more frequently observed in soils with K content less than 0.3 cmol c dm -3 (Sale & Campbell, 1986;Serafim et al., 2012).Veiga et al. (2010) evaluated the influence of K fertilization on the quality of soybean seed produced in soil with a mean K content of 0.2 cmol c dm -3 and did not find statistical significant differences.Regardless K rate applied Pedroso Neto and Rezende (2005), reported grain yield, protein, oil contents, and seed vigor were altered.These results agree with those observed by Snyder and Ashlock (1996) who mention seed quality may be impaired by K soil deficiency.
It is widely reported that K fertilization may influence the quality of soybean seeds, although the results are controversial due to exchangeable K in soil and environmental conditions of each region.Therefore, this study was based on the principle that soybean plants whose soil K is available are more tolerant to deleterious effects (biotic and abiotic) during their life cycle and will produce higher quality seeds and resistance, with an increase in crop yield.In this context, the mean goal of this study was to evaluate changes in the soybean seed quality in two growing seasons after the application of increasing K rates.In addition, we investigated the effects of K fertilization in soybean sowing with adequate levels of soil fertility.

Materials and Methods
This study was carried out in the experimental field of Technology Diffusion Unit (TDU) of Cooperativa Agroindustrial de Maringá (Cocamar) near Floresta, Paraná State (latitude 23º35′42″ S, longitude 52º04′02″ W).The soil of the experimental area was classified as Oxisol.The field study was under no-tillage system during past 20 years, with successive crops of soybean and maize.The climate was classified as Cfa (Alvares et al., 2013).
The experiment was conducted out for two consecutive years with soybean in the 2016/2017 and 2017/2018 growing seasons.The studied treatments tested combinations of the following factors: K rates (0, 40, 80, 120, and 160 kg ha -1 K), and sowing fertilization (0 and 30 kg ha -1 K), composing a fully crossed factorial design (5 × 2), outlined in randomized blocks with four replications.Potassium chloride (KCl) with insurance 58% of K 2 O was used as fertilizer.
Soybean cultivar NA5909 RG was sown at a density of 14 seeds m -1 .The plots consisted of 10 rows, 8 m long, spaced 0.45 m apart.Excepting K fertilization, all cultural treatments from soybean sowing to harvest were carried out according to the guidelines for technologies, products and services (TPS) of Embrapa (2013), and according to the region and the soil chemical and physical analyses.The chemical and physical soil properties were: total organic C 23.2 g dm -3 (Walkley-Black); pH (H 2 O) (soil:water ratio of 1:2.5) 5.55; 15.3 mg dm -3 P and 0.21 cmol c dm -3 K + (both extracted by Mehlich -1 ); 0.0 cmol c dm -3 Al 3+ , 6.4 cmol c dm -3 Ca 2+ , and 1.4 cmol c dm -3 Mg 2+ (both extracted by KCl 1 mol L -1 ); CEC (cation exchange capacity) pH 7 13.6 cmol c dm -3 ; base saturation (%BS) 58%; sand 175 g kg -1 , silt 65 g kg -1 , and clay 760 g kg -1 .K rates were applied manually performed on the soil surface at the V 3 growth stage (Fehr & Caviness, 1977) according to Pauletti and Motta (2017).The sowing fertilization was carried out along the sowing lines with a seeding machine coupled to a tractor.The fertilizer (KCl) was placed at a distance of 0.05 m beside and 0.05 m below the seeds to avoid undesired saline effects during seed germination (Mortele et al., 2009).
Harvesting was performed manually at the R 8 growth stage (Fehr & Caviness, 1977).The yield per 9 m 2 area was determined, the three external rows and 1.5 m at either end of the central rows were disregarded; only the four central rows of the experimental units were harvested.The thousand mass seed was determined according Brasil (2009) weighing eight replicates of 100 seeds, with a moisture correction of 130 g of water per kg of seed.
The electrical conductivity test was measured from 50 seeds per plot previously weighed and placed to soak in 180 mL plastic cups containing 75 mL of deionized water for a period of 24 hours at a constant temperature of 25 ºC (Brandão Jr. et al., 1997).After this incubation, it was possible to determine the electrical conductivity of the solution with Digimed CD 21 conductivity meter, its results expressed in µS cm -1 g -1 .For the K leaching test, 50 seeds per plot were weighed and then immersed in 75 mL deionized water in plastic cups and placed in a germinator at a constant temperature of 30 ºC for 30 minutes.The cups were shaken before and after the time in the germinator, after about 30 mL were removed for determination of K using flame photometry Micronal B-462, its results were expressed of mg of leached K per kg of seed (mg kg -1 ).Both tests were performed in duplicates in each experimental plot.
The water content was determined by the forced air circulation drying oven method, at a constant temperature of 105±3 °C, for a period of 24 hours.25 g of seeds were weighed before placing them into the drying oven and weighed again after removing them from the drying oven (Brasil, 2009).The results were expressed as percent of water (%).
The seed density was determined according to Kryzanowski (2016).It was used a cylindrical vessel with 160.85 cm³, which, when filled, the contents were leveled and compacted by means of three beats of the cylindrical vessel on a rigid surface.After that, the mass was measured and results expressed in g dm -3 .
To determine oil and protein seed contents, the seeds harvested were previously oven dried with forced circulation at 60±1 ºC until obtaining a constant mass in order to standardize the water content.Later, they were milled in a Willey type mill, obtaining the soybean meal with the husk.
Protein amount was determined by digestion of nitrogenous components in presence of heated concentrated H 2 SO 4 , together with a catalytic mixture (copper sulphate and selenium powder) according to the Kjeldahl Semi-micro method.To calculate the conversion of total nitrogen to proteins, factor 6.25 was used and the protein percentage was obtained based on the dry matter.The total oil content was extracted from the Soxhlet extractor apparatus and petroleum ether as solvent, with reflux of 6 hours, in which 2 sub-samples of 2 g from soybean meal were evaluated.The results were expressed as percentage of extracted oil determined by weighing difference.
All data of the variables were subjected to an analysis of basic statistic assumptions using the Shapiro-Wilk (error normality) and Bartlett (homogeneity of variances) tests (p > 0.01).The two growing seasons and effects of the K rates, and sowing fertilization, as well as the possible interactions among the factors were evaluated together by the F test in the analysis of variance.The quantitative data referred to as rates were analyzed by means of regression, and beta coefficients subjected to the t-test, both in relation to the isolated factors and the possible rate interactions.All seed quality variables were subjected to Pearson's linear correlation analysis with soybean yield.For all statistical interpretations, a 5 % probability (p < 0.05) was used (Zimmermann, 2014).

Electrical Conductivity, Potassium Leaching and Potassium Seed Content
For all the variables of quality and seed production analyzed there were differences between growing seasons.In Figure 2 (A, B and C) are the results referring to variables electrical conductivity, K leaching and K seed content.Nonetheless, no interactions among treatments and growing season were shown.Seeds produced in the 2016/17 showed higher averages for electrical conductivity and K leaching when compared to the 2017/18.On the other hand, K seed content was higher for seeds produced in 2017/18.These differences between the growing seasons may be associated with climatic events between the two growing seasons, mainly the rainfall patterns.In the 2016/17 growing season, the cumulative rainfall was 715 mm, while in the 2017/18 growing season, the total was 1097 mm (Figure 1).Furthermore, another factor contributing to this difference it was the storage period of the seeds of the 2016/17 growing season.The seeds harvested were stored until were submitted to the analyzes.Smaniotto et al. (2014)  In the 2017/18 growing season there were no differences for the seed mass, and the average obtained was 140.6 g.One of the possible causes may have been the lower water availability in this growing season, since some authors indicated that the plant responses to available K are clearer in adverse conditions mainly in lower water availability (Sangakkara et al., 2001;Wang et al., 2013;Zörb et al., 2014;Esper Neto et al., 2018).
A significant difference was also pointed in seed yield between growing seasons.However, only in the 2016/17 growing season was affected by K rate (Figure 4C) and interactions among the factors were not assessed.The linear regression model was adjusted, in which each kg of K applied to the soil increased yield in 2.51 kg of soybean seeds.On the other hand, in 2017/18 growing season there were no statically differences and the general average was 3617 kg ha -1 , which corroborates the greater responses of the crops to K fertilization under unfavorable conditions like lower rainfall.
Table 1 shows the averages of all the response variables separated by crop season for the sowing fertilization factor.There were no significant statistical differences among application of sowing fertilization with 30 kg ha -1 of K, and the non-performance of this practice, both for seed quality tests and for components of seed yield.µS cm -1 g -1 mg kg -1 g kg -1 ---------------% --------------g dm

Correlations
Table 2 shows Pearson's correlation analysis.There was no significance for any variable, with oil content.The highest correlations with yield were YIE × EC (r = -0.72) in which the higher seed electrical conductivity, the lower the yield.Castro et al. (2017) found lower values of electrical conductivity in the places where presented better physiological quality of soybean seeds, which were attested by other tests such as germination, vigor and emergence.
Another correlation that stood out was between YIE × DEN (r = 0.70) which is justifiable since density is a component of production.According to Fonseca (2007) plants well-nourished during their development present higher density or larger size of the seed.

Conclusions
Seed quality is greatly affected by higher K in soil as well as K fertilization tends to be helpful to enhance seed quality under low water availability.
The response of seeds oil content was larger than protein content after K fertilization.The sowing fertilization with 30 kg ha -1 had no influence on seeds the quality produced.
The K leaching test and electrical conductivity despite having the same principles did not complement it selves in the situation of this study, since the fertilization increased the absolute seed K contents.Therefore, the electrical conductivity test was more reliable than K leaching test.
Despite soil K levels suitable for soybeans before the experiment, seed yield and mass increased after K fertilization for 2016/17 growing season.Accordingly, K fertilization could be a pathway to reach high seed quality in soybean crop production.
Note.SF = Seeding fertilization; WSF = Without seeding fertilization; YIE = soybean yield; EC = electrical conductivity; KLEA = Potassium leaching; KC = potassium content; PTN = Protein content; OIL = oil content; WC = water content; DEN = seed density; TMS = thousand mass seed.Averages followed by the same capital letter in the column, do not differ from each other, 5% of probability by the F test.
Filho (2012)ed that different storage conditions such as time, temperature and environment can significantly alter quality tests, such as electrical conductivity.BatistellaFilho (2012)applying K rates cited that water unbalance could negatively influence, especially when it occurs near the harvest.

Table 1 .
Means of seed quality and yield for the sowing fertilization factor

Table 2 .
Simple linear correlation matrix between soybean yield and the qualities seed parameters