Introduction
Soybean [Glycine max (L.) Merr.] is one of the most important crops globally due to its high-value grains used as a source of protein and oil for both animal and human consumption (Pagano & Miransari, 2016). Due to the increasing demand for this crop worldwide, the planting area needed is continuously expanding, and alternatives to improve productivity are very active research areas (Masuda & Goldsmith, 2009). According to the Paraguayan chamber of cereal and oilseed exporters (Cámara Paraguaya de Exportadores y Comercializadores de Cereales y Oleaginosas, 2020), Paraguay is the fifth-largest producer and fourth-largest soybean exporter in the world with a production of 10,000,000 t of grains in 3,500,000 ha with a yield of 2,857 kg ha-1. Alto Parana, Canindeyu, Itapua, Caaguazu, and San Pedro are the most productive soybean areas in the country (Cohener & Aguayo, 2009).
The profitability of soybean allows many farmers to invest in improving their production either by increasing the growing area, improving land productivity, or investing in a second planting season (Teixeira et al., 2016). However, the increment of the production area is currently limited due to environmental protection laws restricting the expansion of the crop into new regions (Palau et al., 2012).
Therefore, increasing land productivity is a more practical alternative. This increase in productivity can be achieved with practices that preserve the soil's physical and chemical conditions and allow better pest, weed, and disease control (Gudelj et al., 2018). One of these practices is the use of improved and environmentally specific cultivars (Peluzio et al., 2005; Pires et al., 2005). Additionally, the use of specific cultivars and the adjustment of planting dates and densities can also improve the yield in small areas (Vega & Andrade, 2000).
Thanks to the success of soybean breeding programs, a high number of cultivars on the market are highly productive, resistant to pests and diseases, and adapted to various edaphoclimatic conditions (Sediyama, 2009). The growth of these cultivars is influenced by environmental factors like temperature, rain, relative humidity, soil humidity, and photoperiod. Consequently, the planting season has a decisive influence on the production's quantity and quality (Motta et al., 2000). Therefore, evaluating new cultivars must be a constant practice to provide valuable information for extension agents, consultants, and farmers (Vernetti & Vernetti Júnior, 2009).
Plant density is another factor that can be modified to obtain higher land productivity. Optimal plant density is defined as the minimum number of plants that allow the cultivar to achieve its maximum yield (Vega & Andrade, 2000). One of the main concerns among farmers is reducing the amount of seed used per ha to lower the cost of inputs. Thus, understanding the development of different soybean cultivars planted at different planting densities is fundamental for recommendations for the most appropriate guidelines to maximize yields.
Research on adapted cultivars is of fundamental importance to optimize soybean production. In Paraguay, few published works report adapted cultivars that allow optimizing land productivity. So, there is a need for updated information about the eco-physiological behavior and yield of the new varieties offered on the market. Besides, private companies own the existing data, and it is not publicly available. Therefore, the present study aims to establish a benchmark for an appropriate choice of soybean cultivars in Paraguay, evaluating the productive performance of 14 commercial cultivars planted at three plant densities and in two planting seasons in the Yguazu region during 2017-2018.
Materials and methods
Trials were conducted in the experimental field of the Centro Tecnológico Agropecuario del Paraguay (CETAPAR), located in Yguazu, Alto Parana, Paraguay (25º27’41.97’’ S, 55º02’26.66’’ W, and 258 m a.s.l.). We obtained data on rainfall and maximum, minimum, and mean daily temperaturas during soybean cultivation from September 2017 to April 2018 from the meteorological station of CETAPAR. The water balance graph was constructed using the potential evapotranspiration (ETc) data of the experimental área from the MOD16A2 MODIS/Terra net evapotranspiration database with a spatial resolution of 500 m (Fig. 1) (Running et al., 2017). Mean temperatures of 24.9°C and total precipitation of 1,925 mm were recorded. The experimental area's soil was characterized as 69.0% clay, 2.2% organic matter, a pH of 5.9, and a base saturation of 70.3%.
The experiment was arranged using a completely randomized block design with a 14x3x2 factorial arrangement and three replicates. Factor A consisted of 14 soybean cultivars (Tab. 1); factor B consisted of three planting densities (177,700, 266,600 and 355,500 plants ha), and factor C consisted of early (September 20,2017) and late (November 20, 2017) planting seasons. Date selection follows soybean planting practices in Paraguay, which generally begin in September. However, when the weather is not appropriate (excessive rains or lack of rain), producers start sowing in mid-October and, exceptionally, in November. In Paraguay, it is common to carry out a second planting no later than February. Our experimental unit consisted of five rows 5 m long and a space between the rows of 0.45 m (a total of 2.25 m wide). The distance between experimental units within each block was 1 m and between blocks was 3 m. In each experimental unit, the useful plot was delimited at 4.05 m2. To delineate the plots, 1 m was removed from the ends of each experimental plot, and a row from each edge was discarded. Sowing was carried out with a tractor/planter set (model SHP 249, Semeato Plantio Direto, Passo Fundo, RS, Brazil), using a zero-tillage system with a 0.45 cm space between the rows. Other cultural management and treatments followed standard agronomic recommendations for soybean cultivation (Díaz-Zorita & Duarte, 2004). Base fertilization dosage was 150 kg ha-1 of a N-P-K fertilizer (04-30-10) at the time of planting on both planting dates (early and late planting), in order to meet the nutritional needs of the crop.
When the crop reached the phenological stage of harvest maturity (R8) (Fehr et al., 1971), manual harvesting was carried out from the useful plot of each experimental unit. The number of pods per plant (NPP) was quantified by taking 10 plants per experimental unit, and the average NPP was calculated for each experimental unit. To obtain yield and 1000-grain weight (TGW) data, plants of the useful area were threshed. Weight was determined on an electronic balance (AJ150, Mettler Toledo, Columbus, Ohio, USA), and the value was divided by the harvested area and later extrapolated to kg ha-1. Subsequently, the TGW was determined by quantifying four subsamples of 1000-grains for each experimental unit with a seed counter (KC-10, Fujiwara®, Seisakusho, Japan), and the seeds were weighted with a digital precision balance (JA2003, Hongzuan, Shangai, China). The TGW and yield were adjusted to 13% humidity. The number of grains per m-2 (NG) was estimated from the ratio of yield to seed weight.
Data analysis
The effect of cultivars, plant densities, planting season and their interaction on the yield, NG, TGW, and NPP were studied. For statistical analysis, the SAS (version 9.4) and Infostat (version 2017) software were used. The variance analysis (ANOVA) was performed following the instructions described in SAS for completely randomized block designs. The Tukey's significance test with a family-wise error rate of 5% was used for the comparison of treatment means. Equation 1 describes the model used:
where y ijkl corresponds to the variable response of the i-th cultivar, k-th plant density, l-th planting season in the j-th block. This is the general mean of the response variable. V¡ is the effect of the i-th cultivar; B, is the effect of the j-th block; D k is the effect of k-th plant density; (VD) ik is the effect of the ik-th interaction; E l is the effect of the il-th planting season; (VE) a is the effect of the ill-th interaction; (DE) k ¡ is the effect of the kl-th interaction; (VDE) ikl is the effect of ikl-th interaction; and e ijkl is the experimental error.
A principal component analysis (PCA) of the agronomic variables of soybean cultivars was performed using data from the variables NG, TGW, NPP and the yield. Results from the PCA are shown as a biplot to illustrate the correlation between the variables.
Results and discussion
The effects of experimental factors on the response variables evaluated are summarized in Table 2. The effect of the blocks was not significant for any of the response variables. There was statistical evidence of second-order interactions only for the effects of the cultivarxplanting season interaction on the variables yield, NG, and TGW (P<0.0001). A significant third-order interaction was observed for NPP on the effects of the cultivarxdensityxplanting season interaction (P<0.0001). The simple result of the density was only effective for the variable yield (P = 0.0002).
The factor cultivar consisted of 14 soybean cultivars (Tab. 1), density consisted of three planting densities (177,700; 266,600, and 355,500 plants ha-1), and planting season consisted of early (September 20, 2017) and late (November 20, 2017) planting seasons.
Yield
No significant interaction was observed between the effects of the factors cultivar, density, and planting season (P = 0.4286), not even when considering the interactions between densityxplanting season (P = 0.1645) or cultivarxdensity (P = 0.8937). However, the interaction between cultivarxplanting season was highly significant (P<0.0001). This interaction implies that the yield of some cultivars is not affected by the planting season, while in others, the yield increases or decreases significantly depending on the planting season. A similar yield was observed during both planting seasons for cultivars 5907-IPRO, NA-5909-RG, and DM-5958. On the other hand, the yields of cultivars DM-6563-IPRO, TMG-7062-IPRO, 6505-B, NA-5909, M-6410-IPRO, DM-6262-IPRO, SOJAPAR-R19, 6806-IPRO, 6205-B,M-5947-IPRO, and SYN-1163-RR were significantly higher for the early planting season compared to the late planting season (Fig. 2). Therefore, we can infer that water scarcity affected yields since there was a more significant water deficit in the last season than during the first (Fig. 1). Low yields may be the result of water stress at a critical phenological time. The adverse effects of the lack of water are particularly evident during flowering, seed formation, and seed filling. Lack of available water can reduce yield by reducing the number of pods, the number of seeds, and the mass of seeds that corresponds well with our data (Desclaux et al, 2000).
Because the plant density factor did not interact with any other experimental factor, its simple effect on yield was analyzed (P = 0.0002). The plant density that achieved the highest yield was 266,666 plants ha-1. On the other hand, the lower seeding density produced significantly lower yields than the densities of 266,600 and 355,500 plants ha-1. However, the yield for these last two densities was not statistically different (Fig. 3).
Similarly to the results obtained in this research, the reduction of soybean yield due to late plantings has been reported in previous studies (Giron et al, 2014; Martignone et al, 2016; Teixeira et al, 2016). The planting delay harms yield due to the influence of a smaller number of daily light hours, lower precipitation, and the high temperatures to which the plants are subjected during their initial phase (Martignone et al, 2016). These factors lead to a shorter duration of the vegetative stage, a lower number of nodes per plant and leaf area index, and less dry matter accumulation. Also, the canopy's delayed and inefficient closure causes a more significant loss of water by evaporation (Toledo, 2019).
The soybean response to plant density variations depends on the genotype, soil water conditions, and geographic location (Gaso, 2018). In most soybean cultivars, the response to higher plant density is hindered due to the ability to compensate for gaps between plants, generating longer branches and reducing the energy use for grain filling (Cox & Cherney, 2011). However, different authors mention that soybean yields do not increase significantly when plant densities range from 100,000 to 600,000 plants ha-1 (Lee et al, 2008; Thompson et al, 2015).
According to Rodriguez et al (2015), at densities lower than 200,000 plants ha-1, there is no competition between plants and the number of branches and pods per plant increases; but the tradeoff fails to compensate for the lower number of plants. Therefore, yield is reduced, a situation that was observed at a density of 177,700 plants ha-1. Gaso (2018) found that a significant increase in yield is observed using a density of300,000 plants ha-1 and, above this density, yields do not increase. In this research, we observed that yields did not increase significantly above 266,600 plants ha-1.
Increasing plant density maybe beneficial to mitigate the adverse effects of planting delay that would allow a better use of resources through maximum soil coverage and minimal water loss (Toledo, 2019). Higher plant densities compensate for spaces not covered by the canopy, increasing the number of nodes per m-2 (Martignone et al, 2016).
Number of grains
For the number of grains per m-2, no statistical evidence of significant interaction was observed between the cultivar, density, and planting season effects (P = 0.3688) (Tab. 2). No interaction was observed between the density and planting season factors (P = 0.8608) or between the cultivar and density (P = 0.8937). However, the interaction between cultivar and planting season was highly significant (P<0.0001); this implies that the NG that can be produced depends on combining a specific cultivar and the planting season. The cultivars that obtained the same NG in both planting seasons were TMG-7062-IPRO, 5907-IPRO, NS-5959-IPRO, NA-5909-RG, NA-5909, 6262-IPRO, 6205-B, NS-5959-IPRO, M-5947-IPRO, and SYN-1163-RR. On the other hand, the cultivars DM-6563-IPRO, 6505-B, M-6410-IPRO, SOJAPAR-R19, and 6806-IPRO obtained a higher NG in the early planting season compared to the late one (Fig. 4).
The NG was the component that best explains crop productivity variations (Toledo, 2018). The increase of this component is directly proportional to the duration of the period between the emergency and the start of grain filling (R5). The NG is strongly associated with canopy photosynthesis and also the growth rate of the crop during flowering and pod development (growth stage R1-R5) (Egli, 2013). Besides, a particular relationship is implied between the number of nodes per area and the NG. The greater the number of nodes, the greater the NG. This characteristic is related to the cultivar, environment, and management. Thus, to maximize soybean yields, genotypes with a higher number of plant nodes and rapid soil coverage must be selected since they intercept more than 90% of radiation by R5 (Martignone et al, 2016). The decrease in NG observed in most cultivars in the second season maybe because the delay in planting shortened the plant cycle, causing a lower rate of photosynthesis, less growth and, therefore, a reduction in the production of nodes and grains per m-2. In this research, the density of plants did not influence the NG.
However, it can be expected that increasing plant density will typically maximize the NG due to the increase in the number of nodes per m-2 (Gan et al., 2002).
1000-grain weight
The interaction between the cultivar and planting season was highly significant (P<0.0001). The cultivar that obtained a similar weight of 1000 grains in both planting seasons was 6806-IPRO (Fig. 5). The rest of the cultivars had a significantly higher TGW during the early planting season than during the late planting season.
Grain weight is the second component that best explains soybean yield and is an inherent characteristic of the cultivar (Toledo, 2018). The water deficit during November (Fig. 1) could have influenced the decrease in TWG in the second sowing season. Moreover, delayed planting causes a lower daily accumulation of dry matter during the reproductive stage. The TGW tends to be lower as temperature and solar radiation decrease, resulting in the interruption of grain filling as autumn approaches (Martignone et al., 2016; Teixeira et al, 2016).
Number of pods per plant
The interaction between the effects of the factors cultivar, density, and planting time affected the number of pods per plant (P<0.0001). This implies that the number of pods that the soybean plant can produce will depend on the cultivar, planting density, and planting season. The cultivars that had a NPP higher than 80 were SOJAPAR-R19, SYN-1163-RR, 6806-IPRO, M-6410-IPRO, DM-6262-IPRO, 6505-B, NA-5909, DM-5958. However, due to the interaction, the effect on the NPP is complex; for example, at a density of 355,500 plants ha-1 cultivar 6505-B produced an average of 30 pods perplant during the late planting season, while in the early planting season the average was 40 pods per plant. The same cultivar with a density of 266,600 plants ha-1 during the late planting season produced an average of 45 pods per plant, while during the early planting season, the average was 86 pods per plant. Similarly, with a density of 177,700 plants, the average number of pods per plant was 58, but with the same density during the early planting season, the average was 80 pods per plant (Fig. 6). In contrast, other cultivars such as M-5947-PRO did not significantly increase NPP regardless of the density and planting time. In general, low densities during the first planting season allowed obtaining a higher NPP with the SOJAPAR-R19 cultivar at a density of 177,700 plants during the early planting season, showing the highest NPP (121 pods per plant).
The formation of pods begins in the phenological phase R3 and ends in R6. Pod development is delayed at temperatures below 22°C and tend to fall from the plant with long photoperiods and temperatures greater than 32°C (Toledo, 2018). The formation of pods is susceptible to various types of stress, such as water deficit or the presence of pests and diseases (Toledo, 2019). Moreover, the quality and quantity of solar radiation that reaches the lower layers of the canopy stimulate the establishment of reproductive structures in soybeans (Quijano & Morandi, 2011). The delay of planting causes the canopy to take longer to close, explaining the higher NPP in some genotypes in the second planting season.
The NPP tends to increase with lower plant densities (Toledo, 2019). The number of pods of branches and stems and the number of branches per plant are strongly associated with phenotypic plasticity of indeterminate soybean cultivars. Soybean growth then compensates a lower plant density by showing higher branch emission, higher branch growth and stems, and higher NPP (Balbinot Júnior et al, 2018). Therefore, a higher NPP was observed at a lower plant density.
Principal component analysis
The regions in the biplot contained groups of soybean cultivars with similar characteristics. Cultivars that were closely clustered in one region of the plot represent cultivars that have similar performance patterns. Vectors pointing roughly in the same direction represented yield components that have positive correlations. Yield was slightly correlated with NPP (r = 0.23, P = 0.0002), while the response variables NG (r = 0.60, P<0.0001) and TGW (r = 0.56,P<0.0001) showed a higher correlation with yield. The TGW was not correlated with NG (r = 0.03; P = 0.598) and was weakly correlated with NPP (r = 0.17, P = 0.003), while NG was weakly correlated with NPP (r = 0.22; P = 0.0005). The principal component (PCI) accounted for 43.4% of the variance. The second component (PC2) accounted for 35.7% of the variance. Together, these components accounted for 79.1% of the variance during experiments. In the biplot, the cultivars that obtained the highest average yield are to the graph's right, while cultivars with the lowest yield are to the left (Fig. 7).
This research provided valuable information on the productivity obtained from the interaction between genotype and environment for Paraguayan conditions. It is essential to consider that the meteorological conditions are different each year and that the performance of these genotypes could vary in different locations.
Because of the abundant supply of soybean cultivars in Paraguay, we suggest that subsequent experiments evaluate the impact of climatic factors during the cycle on the yield and its components for cultivars of different maturation groups at different planting dates. Correspondingly, these experiments should include other soybean growth components, such as the leaf area index and the number of nodes per m-2.
Conclusions
This study highlighted the importance of selecting soybean genotypes according to their response to variations in planting date and plant density to increase crop production in the same area. The interaction between the soybean cultivar and planting season affected soybean yield components. Therefore, specific cultivars should be chosen for the early planting season. Cultivars DM-6563-IPRO, TMG-7062-IPRO, 6505-B, NA-5909-RG, M-6410-IPRO, DM-6262-IPRO, SOJAPAR-R19, 6806-IPRO, 6205-B, M-5947-IPRO and SYN-1163-RR can be recommended for early plantings because they show the highest yields, while for late plantings, we recommend cultivating NS- 5959-IPRO. Cultivars 5907-IPRO and DM-5958 can be planted in both seasons because they show similar yields for the two planting seasons. Significantly higher yields were obtained starting from a density of 266,600 plants ha-1. The cultivarxplanting season interaction affects the weight of 1000-grains that determines the quality of the grain. Therefore, the early planting season provided a greater 1000-grain weight than the late planting season. The number of pods per plant depended on the cultivar, density, and planting season. Still, since it correlated poorly with yield, this characteristic could be less important when selecting a cultivar. The combination of the number of grains and the 1000-grain weight had a more significant influence on the generation of yield in the soybean cultivars evaluated. Therefore, it can work as a proxy of yield to select new cultivars.