Structural analysis of soybean pods and seeds subjected to weathering deterioration in pre-harvest estrutural de vagens e sementes de soja submetidas à deterioração por

– The objective of this work was to analyze structurally the pods and seed coats of soybean ( Glycine max ) cultivars, as well as to determine the quality of seeds when subjected to deterioration by weathering in pre-harvest. A 7×3 factorial arrangement was used – seven cultivars and three volumes of simulated rainfall (0, 54, and 162 mm). Exposure to rainfall was simulated in the R8 phenological stage. Then, the plants were taken to a greenhouse, where they were kept until the time of collection of pods and seeds. After collection, the following evaluations were carried out: tetrazolium test, lignin content in pods and seed coats, and structural analysis of pods (exocarp, mesocarp, and endocarp thickness) and seed coats (epidermis, hypodermis, and parenchymal-cell thickness). Pre-harvest rainfall of 54 and 162 mm reduces the quality of soybean seeds; however, the response to deterioration by weathering differs according to the cultivar. The tolerance to all simulated rainfall was greater for BRSMT Pintado, BRS Jiripoca, and M8210IPRO and lower for BRS 1010IPRO. Pods with a greater thickness of the exocarp, mesocarp, and endocarp and a high lignin content show greater resistance to weathering deterioration and seeds with greater vigor and viability. The thickness of the hypodermis of the testa is related to resistance to weathering deterioration and to the obtainment of high-quality


Introduction
Seeds are the basic and vital input -including technologies introduced by breeding -for a sustainable increase in yield and agricultural production (Sharma et al., 2015). Therefore, breeding programs should consider the seed quality factor.
The quality of seeds depends on the seed production process, adequate sowing dates, and suitable areas for production. To meet crop requirements, seed maturity should occur under milder temperatures and low rainfall volumes and the most critical phenological phases should coincide with periods of favorable meteorological conditions (Wang et al., 2012;França-Neto et al., 2016). However, these conditions are not always present, especially in tropical regions where seed placement at sowing, even in seed-producing areas, is usually done to make the cultivation of the second crop viable.
In the case of soybean [Glycine max (L.) Merr.] development, high volumes of rainfall in certain periods, mainly in pre-harvest, can be harmful to seed quality because of weathering deterioration (Pinheiro et al., 2021). This damage causes wrinkling and splitting in the seed coat due to the expansion and contraction of this tissue when exposed to alternating cycles of temperature and relative humidity (Forti et al., 2013;França-Neto et al., 2016;Monteiro et al., 2021).
The susceptibility to weathering deterioration is related to the morphological characteristics of the soybean seeds, since the vital parts of the embryo are situated under a seed coat that is not very thick and that practically does not provide any protection (França-Neto & Henning, 1984). In addition, with successive hydrations, the seed coat begins to change physiologically and structurally (França-Neto et al., 2016). Therefore, anatomical and physiological studies related to the integrity of the pod and seed layers can assist in understanding the response of genotypes to weathering damage, as well as in identifying characteristics with potential for use in breeding programs.
Permeability is a conditioning factor in resistance to weathering damage and is associated with seed coat structure. According to Ma et al. (2004), the cuticle of the palisade layer is determining for seed coat permeability, which is also dependent on the thickness of the epidermal and hypodermal layers, presence of pigments, and cell shape and organization (Mertz et al., 2009).
Weathering deterioration may cause physiological disturbances as reported by Forti et al. (2013). The authors found that the layers of palisade cells, hourglass cells, and of the parenchyma of soybean seed coats gradually decreased in size in the region opposite the hilum, where the wrinkling caused by weathering deterioration usually occurs. The hourglass cells also had a twisted, folded, and flattened appearance.
Regarding resistance to weathering deterioration, semi-permeability related to lignin content is an important trait. According to Oliveira et al. (2014), the study of pod lignin content may be of great value for seed technology since pods with more lignin may show less permeability to water, leading to a lower deterioration of seeds in the field. Several authors confirmed this result both for pods and seed coats, concluding that lignin content differs among genotypes Huth et al., 2016;Bellaloui et al., 2017) and may be related to weathering deterioration (Castro et al., 2016;Huth et al., 2016) just as other characteristics linked to permeability.
The objective of this work was to analyze structurally pods and seed coats of soybean cultivars, as well as to determine the quality of seeds when subjected to deterioration by weathering in pre-harvest.

Materials and Methods
The experiment was carried out at the Seed Physiology, Technology, and Chemistry Laboratories, in the Seed and Grain Technology Center of Embrapa Soja, as well as in the Plant Anatomy Laboratory of Universidade Estadual de Londrina, both located in the municipality of Londrina, in the state of Paraná, Brazil.
Seeds of the used cultivars were produced under greenhouse conditions, with partial control of temperature and relative humidity, which were monitored during the experiment using the data logger device (Figure 1). Type of technology: I, intact; C, conventional; and RR, roundup ready. Source: based on information available at the cultivar registry of Ministério da Agricultura, Pecuária e Abastecimento (Brasil, 2022  Seeds were sown on 10/27/2015, in 9.0 L pots, at a depth of 3.0 to 5.0 cm, in a soil classified as a Latossolo Vermelho Eutroférrico (Santos et al., 2018) of clayey texture, i.e., an Oxisol, with soil amendments according to crop needs (Tecnologias…, 2013). Hours before sowing, the seeds were inoculated with the BIAGRO NG commercial liquid inoculant (Bayer S.A., Cambé, PR, Brazil), containing the SEMIA 5079 and SEMIA 5080 Bradyrhizobium japonicum bacterial strains (5×10 9 viable cells per milliliter), at a rate of 100 mL of product for each 50 kg -1 of seeds. Seeds were treated with the Derosal Plus commercial fungicide (Bayer  Four replicates of five pots with two plants, totaling 40 plants, were used for each treatment. Plants were irrigated daily using drip nozzles. Insecticides and fungicides were applied according to the needs of the crop and to the recommendations for its cultivation (Tecnologias…, 2013).
Rain simulation was performed in the R8 crop development stage (full maturity with 95% of the pods with mature color) during three consecutive days in order to obtain the mean daily rainfall and accumulated rainfall (Table 1); water depth was measured using rainfall gauges distributed across the entire experimental area. The phenological stage of the plants was determined based on the scale of Ritchie et al. (1997).
Rainfall was simulated, at different depths, in a closed room, using a projected device with flat jet spray nozzles to produce large droplets. The device consisted of a 3.0 m high metallic structure to which a suspended "wheeled support" can be connected at a height of 2.5 m. This support has a spray bar for the rain simulation system to move across an area of 15 m 2 in the direction of the length of the device. The bar of the device was pulled by chains and gears with the assistance of an electric motor adjusted by a frequency modulator, allowing a previously determined constant speed.
A hydraulic pump, with constant pressure and automatic activation, was used to pump stored water from a 3,000 L capacity tank to the spray bar equipped with the nozzles responsible for forming the simulated raindrops. A total of seven TKSS20 high-flow conical nozzles (Spraying Systems do Brasil, São Bernardo do Campo, SP, Brazil), spaced at 0.50 m, were used to provide a greater rainfall uniformity over the area. Operating pressure was 0.81 kgf cm -2 , the height of the bar was 1.45 m in relation to the surface of the experimental units, and the displacement speed was 0.050 m s -1 . Therefore, for each total displacement of the bar, 0.9 mm of simulated rainfall was applied. These specifications led to the production of artificial raindrops with a mean volumetric diameter of approximately 1,140 micra, according to information from the manufacturer of the nozzles (Spraying Systems do Brasil, São Bernardo do Campo, SP, Brazil).
After the hydration and dehydration cycles, the plants were taken back to the greenhouse, where they were kept until the time of collection of pods and seeds (time of harvest), within three to four days. Pods and seeds were collected from the upper, middle, and lower third of the plant in equal proportion. After collected, the pods and seeds were sent to the laboratory for analyses, which included the tetrazolium test, structural analysis of pods and seed coats, and determination of lignin content of pods and seed coats.
For the tetrazolium test, two subsamples of 50 seeds per replicate were used, being pre-conditioned on germination test paper that was moistened with distilled water for 16 hours in a seed germinator at 25ºC. After this period, seeds were totally submerged in tetrazolium solution (2,3,5-triphenyl tetrazolium chloride) at a concentration of 0.075% and kept at 40°C for approximately 150 min in a germination chamber in the absence of light. Then, the seeds were classified according to the criteria -viability, vigor, weathering damage (classes 1-8 and 6-8) -proposed by França-Neto & Krzyzanowski (2018). The results were expressed in percentage.
For the structural analysis of the pods and of the seed coats, the samples were initially fixed in FAA 50 medium, consisting of formaldehyde, glacial acetic acid, and 50% ethyl alcohol at a ratio of 1:1:8 (Johansen, 1940). To visualize the structural differences between pods and seed coats, the following steps were taken. The pod and seed coat fragments were fixed in paraffin. The EM UC7 ultramicrotome (Leica Microsystems GmbH, Wetzlar, HE, Germany) was used to slice cross sections with 1.0 μm thickness in the region opposite the connection to the funiculus of the pod and opposite the seed hilum. The plant tissue was then stained with 1.0% Astra Blue and 1.0% fuchsine and visualized in the BX 51 OLYMPUS optical microscope (Olympus Corporation of the Americas, Center Valley, PA, USA), at 40x magnification. The images were digitalized by a video camera adapted to the microscope and processed by a microcomputer using the Motic Images Plus 3.0 software (MoticEurope, S.L.U., Barcelona, Spain). The results for thickness of layers of the pods (exocarp, mesocarp, and endocarp) and seed coats (epidermis of the testa, hypodermis of the testa, and parenchymal cells) were presented in µm.
To determine the lignin content of the pods and seed coats, four replicates of 100 seeds and 50 pods Pesq. agropec. bras., Brasília, v.57, e02697, 2022 DOI: 10.1590/S1678-3921.pab2022.v57.02697 were used for each treatment. Initially, the seeds were separated from the pods and then immersed in water for 12 hours to separate seed coats from cotyledons. After this procedure, the seed coats and the pods were dried in a laboratory oven, at 105°C, for 24 hours. The dry matter obtained was ground and homogenized. After that, 0.3 g was weighed for the extraction of proteins connected to the cell wall. After the material free of proteins was obtained, the amount of lignin was quantified by the acetyl bromide method (Moreira-Vilar et al., 2014). The obtained data were analyzed for normality and homoscedasticity using the Shapiro-Wilk and Hartley tests, respectively, which indicated that data transformation was not necessary. The analysis of variance was performed, and means were compared by the Scott-Knott test, at 5% probability. Analyses were carried out using the SISVAR computer software (Ferreira, 2011).

Results and Discussion
The summary of the analysis of variance for the effects of cultivars and simulated rainfall volumes for the evaluated variables, as well as the mean squares for the isolated effect of cultivar for lignin content, are presented in Table 2.
Data on weathering damage determined by the tetrazolium test showed that rainfall simulations were effective in differentiating the cultivars using the three volumes initially proposed (Table 3). This made it possible to determine the genotypes susceptible or resistant to deterioration in pre-harvest, as well as to verify the effect of these factors on the structural composition of pods and seed coats. It is important to highlight that the maximum and minimum simulated rainfall volumes corresponded to the annual historical mean values found for the pre-harvest period of the main soybean-producing regions of Brazil (Inpe, 2017). For this reason, the obtained results show the similarity between the tested variables and the reality observed by the seed grower in the field.
The structural differences among the pods and seed coats of the tested cultivars were also clearly visualized by morphological characterization, though optical microscopy. It was possible to identify the structures of the pericarp of the pods, for example (Figure 2 A). According to Appezzato-da-Glória & Carmello- Table 2. Summary of the table of analysis of variance for the data obtained both for the structural analysis and lignin content of pods and seed coats and for the quality of seeds of soybean (Glycine max) cultivars under different volumes of simulated rainfall (0, 54, and 162 mm), in pre-harvest, in 2016, in the municipality of Londrina, in the state of Paraná, Brazil (1) . Guerreiro (2006), the pericarp can be analyzed structurally by delimitating its regions, usually denominated exocarp, mesocarp, and endocarp. For seed coats, different thicknesses were verified for the layers epidermis of the testa, hypodermis of the testa or hourglass cells, and parenchymal cells (Figure 2 B). However, it was not possible to evaluate the cuticle layer associated with the epidermis of the testa. When characterizing seeds from soybean cultivars, Mertz et al. (2009) observed those same structures, as well as contrasting permeability for seed coats; however, the authors were also not able to analyze the cuticle layer. Ma et al. (2004) concluded that the cuticle layer is related to seed coat permeability, which may be weaker or stronger, when showing or not cracks, respectively. This information is important since the cuticle, a fine external layer of the seed coat made up of polymers, such as cutin and cuticular waxes (Appezzato-da-Glória & Carmello-Guerreiro, 2006;Smýkal et al., 2014;Vu et al., 2014), is the first barrier to water imbibition, showing a varying structure according to genetic, physiological, and environmental factors. Therefore, knowledge of the cells that compose pods and seed coats is essential because it is through these layers that water can reach the embryo of the soybean seed.
For the cultivar effect, the characterization of the thickness of the pod layers showed that the exocarp from the BRS 1010IPRO and BRS 284 cultivars was not as thick as the other ones (Table 4). Originating from the outer epidermis of the ovary, the exocarp is the outermost layer of the pod and may have a cuticle layer, stomata, and trichomes; it may also have one or more layers of lignified sclerenchymatous cells, conferring it rigidity (Appezzato-da-Glória & Carmello-Guerreiro, 2006), which directly affects water uptake. Therefore, cultivars with pods with a thinner exocarp may be less tolerant to moisture fluctuations in the environment since water would be rapidly absorbed when crossing a shorter distance, i.e., a thin cell layer. This result was confirmed indirectly by the greater percentage of weathering damage, within classes 1-8 and 6-8, in the seeds from pods with a more limited exocarp thickness (Table 3).
Another factor that may be directly related to water uptake through the exocarp is lignin content (Table 4). According to the obtained results, the pods of cultivars BRS 1010IPRO and BRS 284 contain a lower lignin content than all others. Lignin is formed through the oxidative coupling of monomers of hydroxycinnamyl alcohols, called monolignols, and, besides affecting the degree of resistance to mechanical damage, may also affect the process of water absorption (Zhao & Dixon, 2011;Kuchlan et al., 2018;Liu et al., 2018). This Table 3. Vigor, viability, and weathering damage determined by the tetrazolium test carried out on seeds of soybean (Glycine max) cultivars under different simulated rainfall volumes (0, 54 and 162 mm), in pre-harvest, in 2016, in the municipality of Londrina, in the state of Paraná, Brazil (1) . (1) Mean values followed by equal letters, lowercase in the rows and uppercase in the columns, do not differ from each other by the Scott-Knott test, at 5% probability. effect can be attributed to the hydrophobic character of lignin, which slows the passage of water through tissues, especially when associated with other waxy substances, such as suberin (Vishwanath et al., 2014;Abati et al., 2022). The mesocarp of the pods of BRS 1010IPRO, BRS 284, and NA 5909 RG was not as thick as the other ones (Table 4), as observed for the endocarp of the first two cultivars. The obtained results are indicative that the pod layers were not affected by the volumes of simulated rainfall and that the differences found among cultivars are related to the genetic characteristics of each material. Therefore, cultivars with thicker pods can be used in breeding programs as an indicator for selection of materials resistant to weathering deterioration in pre-harvest both due to the phenotypic stability of their pods and to their qualitative character, resulting in a low genotype by environment interaction.
Seed coat measurements showed an interaction between cultivars and the simulated rainfall volumes for the hypodermis of the testa, but there was no  significant effect of the evaluated factors on the epidermis of the testa and parenchymal cells (Table 2). Cavariani et al. (2009) obtained a similar result when analyzing the thickness of the palisade layer, thickness of the hypodermis, and total thickness of the seed coat of soybean seeds, finding that the applied treatments did not differ significantly in any cultivar × location combination.
For the hypodermis of the testa, the increase in simulated rainfall led to a reduction in the thickness of the hourglass cells (Table 4). Forti et al. (2013) highlighted that the wrinkling caused by weathering damage led to a reduction in the thickness of these cells, which showed a twisted, folded, and flattened appearance.
Regarding the hypodermal layer, a greater thickness was observed for BRS 1010IPRO, BRSMG 752S, BRS Jiripoca, and M8210IPRO in the treatment without rainfall (Table 4), indicating genetic differences between cultivars. For cultivar BRS Jiripoca, at 54 mm of rainfall, the layer maintained its structure under excess weathering, showing a greater tolerance to rainfall than all others. For the maximum simulated rainfall of 162 mm, cultivars BRSMT Pintado and BRS Jiripoca, with a genetic trait of tolerance to rainfall in pre-harvest, showed the best results and BRS 1010IPRO, BRS 284, and NA 5909 RG, the worst ones. This result confirms those obtained in the evaluation of pod structure (exocarp, mesocarp, and endocarp).
No relationship was observed between the hypodermis of the testa and lignin content in the seed coat (Table 4). According to Bahry et al. (2015), water absorption in seeds depends on the conditions in which they were formed, varying among genotypes, regardless of the color of the seed coat and its concentration of lignin.
For vigor, cultivars NA 5909 RG and M8210IPRO exhibited higher values in the treatment without any application of water, i.e., without simulated rainfall (Table 3). At a rainfall volume of 54 mm, the cultivars whose seeds showed the greatest vigor were NA 5909 RG, BRSMT Pintado, BRS Jiripoca, and M8210IPRO. However, at 162 mm, the last three cultivars had a better performance than the first one, with the lowest seed vigor. Since there was no statistical difference for lignin content in the seed coat among cultivars BRS 1010IPRO, BRS 284, BRS Jiripoca, and M8210IPRO, it was not possible to verify the relationship between lignin content in the seed coat and seed vigor, within the limits of contrasts evaluated in the present work.
Overall, increased rainfall volumes reduced the vigor of the tested seeds, except for cultivars BRSMT Pintado, BRS Jiripoca, and M8210IPRO (Table 3). Maintenance of vigor with the increase in rainfall in the pre-harvest period may be a parameter for studies aiming to determine resistance factors and seed quality under weathering deterioration. This attribute is important since high vigor seeds result in a greater percentage, speed, and uniformity of seedling emergence in the field, especially under unfavorable conditions; moreover, high vigor seeds contribute positively to crop yield performance (Tavares et al., 2013;Marcos-Filho, 2015;Bagateli et al., 2019).
The results for viability confirm those obtained for the vigor test. With the maximum rainfall of 162 mm, the cultivars that showed the best results were BRSMT Pintado, BRS Jiripoca, and M8210IPRO, whereas BRS 1010IPRO had the lowest viability for all simulated rainfall volumes (Table 3).
Considering the obtained results, a greater tolerance to fluctuations in ambient moisture was observed for seeds from pods with thicker walls, particularly when associated with a high lignin content, leading to a lower percentage of weathering damage and a greater vigor and viability of seeds. The thickness of the hypodermis layer is also directly related to resistance to weathering damage and to vigor.

Conclusions
1. The simulated pre-harvest rainfall of 54 and 162 mm reduces the quality of soybean (Glycine max) seeds, which differs according to the responses of the used cultivars to deterioration by weathering.
2. The BRSMT Pintado, BRS Jiripoca, and M8210IPRO cultivars have greater tolerance to weathering deterioration in pre-harvest, while BRS 1010IPRO has the lowest seed vigor for all simulated rainfall.
3. Soybean pods with a greater thickness of the exocarp, mesocarp, and endocarp, as well as a high lignin content, have a greater resistance to weathering deterioration and seeds with greater vigor and viability.