The plants used in this work were three tetraploid (2n = 4x = 40) E. curvula genotypes: the sexual OTA-S (USDA: PI574506), the full apomictic Tanganyika (USDA: PI234217) and the facultative apomictic cv. Don Walter. Moreover, a diploid sexual accession, USDA PI208214, was used as a contrasting genotype for the in situ hybridization assays. They were grown in a greenhouse in 10-liter pots, under natural light conditions (Bahía Blanca, Argentina, 38°43’0″S, 62°16’0″O), at 25˚C +/- 4°C. Inflorescences were taken at the beginning of anthesis, when all embryo sac developmental stages are found simultaneously in an individual panicle (Selva et al., 2017).

In order to strengthen the study of the species, a RT-PCR and in situ hybridization were performed with the gene SPL7 (SQUAMOSA PROMOTER BINDING PROTEIN-LIKE7) ( Supplementary Table 2 ), found to be differentially expressed in the full-apomictic and full-sexual genotypes of E. curvula.

We performed a comprehensive confocal laser microscopy analysis and comparison of all the developmental stages of the female gametophyte in three genotypes of Eragrostis curvula: i) the sexual OTA-S, ii) the fully apomictic Tanganyika, and iii) the facultative apomictic Don Walter. As mentioned in the introduction, sexual development comprises the stages of FG0 or MMC (Megaspore Mother Cell), FG1 or FM (Tetrad and Functional Megaspore), FG2 (Two nuclei stage), FG4 (Four nuclei stage) and FG7 (Mature stage). We have used a protocol described by Barrell and Grossniklaus (2013) and modified by Cornaro et al. (2024), which allowed us to distinguish the nuclei (DNA) and the cell walls, and in some cases also the single chromosomes ( Figures 2C2 , 4A ). We focused our analysis on the megasporogenesis and megagametogenesis of the three genotypes, to identify morphological differences. The measurements were organized into three main calculations: the total ovule dimension (height and width) (exemplified in Figures 1A , 2A ), the total embryo sac dimension (height and width) (exemplified in Figures 1A , 2A ). Moreover, the organization of the ovule for the three genotypes was found to be divided in four distinguishable layers in the nucellus of every developmental phase ( Supplementary Figure 2A ), thus, five cells of each layer were measured in the same manner (height and width), to obtain an average dimension of each of the four sections ( Supplementary Figure 2A ). Comparisons were made between the obtained measurements in micrometers, performing a Student’s T-test ( Supplementary Table 1 ).

The ovule is enclosed by the integuments and the development of the embryo sac occurs from the central portion of the ovule. During the initial steps of female development, ovules at pre-meiosis, for all three genotypes it was possible to observe a unique cell in the center of the embryo sac ( Figures 1A–F ), as well as two short immature stigmas attached to the ovule ( Supplementary Figure 2C , MMC).

Regarding the height and width of the total ovule, it was observed that at pre-meiotic stage there were no notable differences in the measurements performed between the three analyzed genotypes ( Figures 1G–J ), where the MMC was clearly distinguished as a bigger nucleus ( Figures 1B, D, F ).

The average ovule height in the sexual genotype OTA was approximately 200µm, similarly to what was found in Don Walter while in Tanganyika was closer to 185µm. The average width was around 180µm in OTA, while it was closer to 165 and 170µm for Tanganyika and Don Walter respectively ( Supplementary Table 1 ). On the other hand, for the embryo sac dimensions, Don Walter’s height (~20 µm) was found to be significantly larger than the fully apomictic Tanganyika (~ 15µm). OTA’s embryo sac height (~ 15.6µm) was closer to Tanganyika’s ( Supplementary Table 1 ).

At this point in the sexual genotype OTA the MMC enters meiosis and forms a reduced tetrad, which was clearly distinguishable – the position of the nuclei of the four megaspores was found to be vertically aligned within the embryo sac ( Figures 1K, L ). The ovule measurements in the tetrad stage in OTA were a height of 282.5µm and a width of 214µm. After the tetrad formation in sexual development, three of the megaspores degenerate and one persists (the one closer to the chalaza), giving rise to the functional megaspore (FM) ( Figures 1L, M ). In FG1 a single nucleus was observed once again, often with a clear degeneration of the megaspores adjacent to it ( Figure 1L ), and it could be distinguished from the MMC stage because of its larger size (average height: MMC ~201µm vs FM ~223µm; average width MMC ~179µm vs FM 224µm) and larger stigmas with the distinctive feathers ( Supplementary Figure 1C ). On the other hand, in the apomictic Tanganika and the facultative Don Walter, the FG1 stage is completely absent, and the ovule enters mitosis directly into FG2. Indeed, the MMC was clearly observed, but a tetrad was never detected in these two genotypes. Concerning the cells in each nucellus layer, there were no significant differences found between genotypes in MMC and FG1 stages ( Supplementary Table 1 ).

At FG2, as the ovule continues growing after the first round of mitosis, the total ovule dimensions obtained revealed that the ovule of the sexual genotype OTA was significantly larger than its apomictic parallels ( Figure 2G , Supplementary Table 1 ), reaching 288µm of height ( Supplementary Table 1 ). With respect to the width, OTA was significantly wider than Don Walter ( Figure 2H ), measuring approximately 270µm. Tanganyika’s width was similar to that of Don Walter’s, being both around 220µm.

Interestingly though, in this stage Tanganyika’s embryo sac width (30 µm) was observed to be significantly bigger than the facultative genotype ( Figure 2J ), with a size comparable to OTA’s embryo sac (22 and 26µm respectively). Concerning the layer cells in the stage of FG2, for the first and second layers, sexual OTA’s cells were found to be longer. Likewise, for the second and third layers OTA’s cells appear to be wider ( Supplementary Table 1 , Supplementary Figure 2B ).

After the second mitosis, at FG4 stage, the four nuclei in all genotypes were observed close together forming a group in the center of the ovule as shown in Figures 2K–P , where the images show three clearly distinguishable nuclei, whereas the fourth one was in all cases in a deeper level tridimensionally.

The ovule and embryo sac sizes were not found to be differential between genotypes for this stage ( Figures 2Q–T ). In OTA the total ovule height was approximately 289µm, while for Tanganyika and Don Walter it was close to 281µm. The total width was ~273µm in OTA, ~238µm in Tanganyika and ~244 µm in the facultative Don Walter ( Supplementary Table 1 ). With respect to the embryo sac, OTA and Tanganyika’s dimensions were more similar between them, while Don Walter showed a tendency to a smaller, rounder embryo sac (DW height ~35 µm and width ~34µm) ( Supplementary Table 1 ).

At this stage, regarding the layer cell analysis, sexual genotype OTA’s cells were observed to be wider for the first, second and third layers, while also being longer for the second and third layers. Nonetheless it was observed that, interestingly, the full apomictic genotype Tanganyika at FG4 showed much longer cells forming the fourth layer ( Supplementary Table 1 , Supplementary Figure 2B ).

Finally, the FG7 octonucleate stage could be detected in the sexual OTA and also in the facultative apomict Don Walter, where we were able to find a sexually developing embryo sac. As it was mentioned in the introduction, in the Eragrostis-type apomictic development, the MMC follows only two rounds of mitosis, generating a tetranucleate embryo sac, so the fully apomictic genotype Tanganyika did not exhibit an FG7 stage. Even though no significant differences were found for the dimensions of the total ovule ( Figures 2Y, Z ), the size of the facultative apomict embryo sac was found to be significantly larger than the sexual genotype ( Figures 2A2, B2 ), with a height of ~114µm and a width of ~93µm ( Supplementary Table 1 ). Surprisingly however, regarding the layer cells, the first, second and third layers were comprised of significantly longer cells in the sexual OTA, while they were wider for the third and fourth layers ( Supplementary Table 1 , Supplementary Figure 2B ).

The identified differential features found to be significant between genotypes are compiled in a schematic summary, exhibited in Figure 5 .

Regarding pollen grain development, microsporogenesis and microgametogenesis were analyzed for the three genotypes in study ( Figure 4 ). The stages observed included the pollen mother cells, meiosis or pre-meiosis, the uninucleate grains and the bicellular pollen. The technique used in the work allowed the obtention of high-quality confocal images of the pollen development and spatial positioning, for the first time in Eragrostis curvula, and it even permitted the observation of chromosomes during the division stage. Nevertheless, it was not found particularly useful for the analysis of the surface texture of the pollen grains. Height and width of pollen grains were analyzed in the four stages observed, but no major differences were found between genotypes.

It was possible to recognize active division taking place in the early stages of pollen development in the sexual genotype OTA, observing metaphase and anaphase with remarkable clarity ( Figure 4A ).

Over the years, the elucidation of the genetic and epigenetic processes controlling ovule development in naturally apomictic species has been considered a key goal, as it would aid enormously in the development of valuable tools for the transference of this trait to economically important species, i.e maize, wheat, and rice (Hanna and Bashaw, 1987). Since asexual reproduction through seeds appears to have arisen more than once during evolution (Carman, 1997), the use of Eragrostis curvula as an experimental apomictic model adds great value to the study, as the complex consists of naturally sexual and apomictic genotypes, even including different ploidy levels. Moreover, this species preserves the normal relationship between the maternal and paternal genome in the endosperm, which has been found to be an important advantage for the transference to major crops and cereals (Carballo et al., 2021). Cyto-embryological studies of apomictic ovules of E. curvula have been performed with optical microscopy, using paraffin-embedded and sectioned material, typically stained with safranin-fast green or using clarification methods to observe pistils under DIC microscopy (Meier et al., 2011; Carballo et al., 2021). During several years, most of these protocols of microscopy imaging have been associated with low resolution and insufficient detail (Hepler and Gunning, 1998), but the introduction of confocal laser scanning microscopy allowed to effortlessly section thick stained samples, enabling the visualization of cells in its tridimensional orientation and providing context of the adjacent tissue, which is a huge advantage in relation to serial slicing (Running et al., 1995).

The characterization and description of ovule and pollen development has been extensively explored on different model species over the years, with the purpose of understanding the mechanisms lying underneath (Schneitz et al., 1995; Kuhn and de Araujo Mariath, 2014; Rossig et al., 2021; Rocha et al., 2023).

The optimized staining technique for whole-mount tissue on which this work was based, proved to be effective in the determination of the stages of megasporogenesis and megagametogenesis also in A. thaliana (Siddiqi et al., 2000; Vieira et al., 2012), as well as in maize (Barrell and Grossniklaus, 2005) and rice (Zeng et al, 2007), allowing the observation of distinct nuclei and cell walls, and overall granting high-quality images with a much less arduous protocol that excludes the time-consuming embedding and sectioning steps. Moreover, measurements using confocal laser microscopy techniques were successfully obtained for different species, such as embryo sac dimensions in rice (Zeng et al., 2007); A. thaliana’s ovule area, cell size and pistil length (Fernandez et al., 2010; Cucinotta et al., 2016; Gomez et al., 2023) and intercellular distances between sperm nuclei and female nuclei in Theobroma cacao (Ford and Wilkinson, 2012), among others. Likewise, in this work it was possible to observe and measure E. curvula’s reproductive tissue in depth, examining all planes of focus and spatial distribution of the structures.

A reproductive calendar of sexual and apomictic individuals of this species was previously presented by Selva et al. (2017) using fixed tissue, but no comparisons were performed between them.

Moreover, most of the reproductive calendars that have been constructed for apomictic grasses correspond to aposporous apomicts (Caceres et al., 1999; Dusi, 2001; Laspina et al., 2008; Sharma et al., 2014; Soliman et al., 2019), where there appears to be a different synchronization between male and female development in the apomictic individuals, which we did not detect during our analysis.

In the developmental description of the grass Brachiaria decumbens (Dusi, 2001), it was found that the aposporous genotypes had a higher growth rate, and that its pistils were larger during anthesis (FG7), even though the comparisons were made between tetraploid and diploid accessions, so this difference could be attributed to the ploidy level. In contrast, in our analyses it was observed that the sexual genotype OTA had a bigger total ovule size and larger dimensions corresponding to the first three layers in study for all stages. On the other hand, the embryo sac dimensions appeared to be larger for the two apomictic E. curvula genotypes for all stages analyzed. Interestingly, regarding the last layer that was observed in the ovule structures ( Supplementary Figure 1 ), the full apomictic genotype Tanganyika exhibited much longer cells, reaching around 13µm ( Supplementary Table 1 , Supplementary Figure 1 ), constituting a potentially important feature for a simple identification of this genotype.

Moreover, it was possible to generate a developmental timeline for the pollen grain development of this species, effectively discerning four key stages in the three contrasting genotypes. Although at this time it was not possible to distinguish differences in shape, size or pollen wall architecture in the different genotypes, this procedure permitted the clear observation of nuclei and chromosomes, even enabling the examination of active division.

In order to strengthen the knowledge about the regulation of the trait apomixis in this species, a transcription factor named SQUAMOSA PROMOTER BINDING PROTEIN-LIKE7 (SPL7) was evaluated in silico and found to be overexpressed in the sexual genotype OTA. To validate this result, an RT-PCR was performed, and the experiment confirmed amplification solely in this genotype, suggesting a role or effect in the regulation of the trait. The assembly of a phylogenetic tree aiming to understand the identity and evolution interactions between the analyzed E. curvula transcript and the sequences corresponding to SPL7 in A. thaliana and SPL9 O. sativa allowed to correctly cluster these genes as having ortholog identity. Subsequently, an in situ hybridization was performed to observe the localization of the expression of the gene, and it revealed a positive signal of expression only on the ovules corresponding to the sexual genotype, particularly in the center of the embryo sac.

The family of SQUAMOSA PROMOTER BINDING PROTEIN-LIKE transcription factors has been linked to reproduction in Boechera (Amiteye et al., 2011) and in a study of Eragrostis curvula’s differentially expressed microRNAs, where some of them were targeting transcripts encoding a SPL protein and a MADS-box transcription factor (Garbus et al., 2019). Particularly, the expression of SPL7 in the grass Panicum virgatum (along with SPL8) was found to be related to inflorescence and stem development, as well as flowering time, and it was observed it acts targeting two transcription factors of the MADS family (Gou et al., 2019). This gene appears to be linked to miR156, and it has also been associated to abiotic stresses in Arabidopsis thaliana and Carya illinoinensis (Stief et al., 2014; Perea-Garcia et al., 2021; Wang et al., 2021), which could be of interest having in mind that a link has been stablished between drought stress and an increase in sexual embryo sacs in facultative apomictic genotypes of E. curvula (Rodrigo et al., 2017; Selva et al., 2020). Interestingly, SPL7 has also been studied as a crucial copper regulator in plants (Yamasaki et al., 2009), and copper homeostasis is suggested to be of extreme importance in plant growth (Schulten et al., 2022), seed production, fertility of pollen (Yan et al., 2017) and also fertility of the gynoecium (Rahmati Ishka and Vatamaniuk, 2020), all of this being closely related to this gene.

Although a direct connection between this gene and apomictic or sexual reproduction has not been confirmed, it can be theorized, in light of the results obtained in this work, that SPL7 is part of a regulatory pathway in sexual development, that could include microRNAs and genes of the family of the MADS-box, and which is repressed or not present in apomictic individuals. Moreover, in A, thaliana, SPL7 was found differentially expressed in anthers and stamen at flower stage 15 (Mergner et al., 2020), meaning that regulation of apomixis through this gene could be triggered by the pollen, such as in other important genes like BBM1 (Khanday et al., 2019).