Microstructure of Croatian Wild Grapevine (Vitis vinifera subsp. sylvestris Gmel Hegi) Pollen Grains Revealed by Scanning Electron Microscopy

Wild grapevine (Vitis vinifera subsp. sylvestris Gmel Hegi) is dioecious with male and female plants, whereas domesticated grapevine is mostly hermaphrodite with self-fertile hermaphrodite flowers. The pollen morphology of wild grapevine has been poorly studied. There is no detailed palynological study of V. sylvestris in Croatia and neighboring countries. Here, scanning electron microscopy (SEM) was used to analyze the pollen of V. sylvestris from male and female individuals growing at two natural sites in Croatia. The selective APT3 marker was used to confirm the flower phenotype with the genetic background. SEM analysis showed that the pollen grains of V. sylvestris were isopolar and radially symmetrical, with foveolate perforated ornamentation, regardless of the flower type of the individuals. All male flowers were 3-colporate and prolate in shape, whereas female individuals varied from subprolate to spheroidal and had inaperturate pollen grains. Pollen shape, dimensions and exine ornamentation proved very informative, and here we address the most polymorphic traits in the analyzed V. sylvestris individuals. Principal component analysis (PCA) and clustering based on pollen morphology variables clearly differentiated individuals by their flower type, and no grouping specific to population was observed, pointing to the conserved pollen structure of V. sylvestris. The results indicate the need to continue the palynological study of V. sylvestris and serve as a good phenotypic basis for functional genetic studies on genes involved in pollen morphology and function.

The wild grapevine is dioecious and presumed to be an ancestor of domesticated grapevines [3]. The crucial difference between the two subspecies is in flower morphology. Domesticated grapevine has mostly hermaphroditic and, in some cultivars, femaletype flowers, whereas wild grapevine exhibits male or female flowers. Recent transcriptomic studies showed that all three types of flowers begin their development as perfect hermaphroditic flowers [4][5][6]. In later stages of flower specification, male flowers show a reduced pistil, style and stigma, whereas females have shorter, backward-reflexed or shriveled stamens with sterile pollen [6,7].

Pollen Microstructure
From the 10 different V. sylvestris genotypes analyzed in this study, eight had functionally male flowers, and two had functionally female flowers ( Figure 1). Regardless of the flower type or sampling location, all pollen grains were isopolar and radially symmetrical, with foveolate-perforate exine ornamentation. Significant differences were observed in the dimensions and shape of the pollen from male and female individuals (Table 1, Figure 2).  Table 1. Morphometric data on pollen grains taken from 10 V. sylvestris genotypes (two female and eight male) from two different natural populations. Each value represents an average measurement  Table 1. Morphometric data on pollen grains taken from 10 V. sylvestris genotypes (two female and eight male) from two different natural populations. Each value represents an average measurement ± standard deviation, based on a minimum of 30 pollen grains per genotype obtained by the scanning electron microscopy (SEM).  [30]. Shape classification according to [30]. Mean values followed by different lowercase letters in columns represent significant differences according to Tukey's test at p ≤ 0.05. The pollen of male V. sylvestris was 3-colporate and prolate (P/E ratio: 1.78-1.93), whereas that of the female was inaperturate: no colpi or apertures, spheroidal to subprolate in shape (P/E ratio: 1.12-1.15).
In this study, the polar axis ranges from~25.0 (female Pak12) to~30.60 µm (male Im19). The equatorial axis ranges from~15.2 µm (male Pak21) to 22.5 µm (female Pak10). The equatorial axis had the lowest significant variability among the studied individuals, but significantly differed in male and female individuals. The mean values for the P/E ratio ranged from 1.12 (female Pak10) to 1.93 (male Pak21).
The number of perforations per µm 2 at the exine surface varied significantly, from 29.67 to 50.67 (both values for male individuals), and two female individuals also had a significant difference in this trait (Pak12 having a significantly higher number of perforations than Pak10). The lowest perforation length (330.14 nm) was observed for male Im11, and the highest for male Pak 11 (506.38 nm). Perforation width was the lowest for Im11 (206.54 nm) and the highest for Im18 (328.86 nm). The highest significant difference in colpi measurements for male individuals was observed between Pak11 and Im19, though colpus width did not vary significantly among the tested genotypes. The mesocolpium width was significantly higher in the case of Pak11 and Pak32 (~10.9 µm, each) compared to Im18 (9.9 µm).  [30]. Shape classification according to [30]. Mean values followed by different lowercase letters in columns represent significant differences according to Tukey's test at p ≤ 0.05.

Groupings of Individuals Based on Pollen Characteristics
PCA analysis (colpi values excluded) (Figure 3a,b) reflected the morphological differentiation represented in Table 1. The first two axes explained 80.94% of the overall variation in the studied set and separated male V. sylvestris with 3-colporate pollen grains on the right side from female individuals with inaperturate pollen grains on the left side of the 2-D plot. Two female individuals (with inaperturate pollen) were distinguished mainly by the equatorial axis length (Figure 3a). The leading phenotypic traits for this sample set were pollen length (P) and grain size-based gradient P/E having a strong but negative correlation with pollen width (E), together explaining 47.82% of the total variation along the first principal component. Exine ornamentation traits (perforation width and length) were strongly correlated to each other along the second principal component, but were independent from pollen dimensions (P, E and P/E) due to a very weak correlation. The number of exine perforations, despite significant differences observed among the genotypes, showed the weakest correlation to other traits (shortest vector).

Groupings of Individuals Based on Pollen Characteristics
PCA analysis (colpi values excluded) (Figure 3a,b) reflected the morphological differentiation represented in Table 1. The first two axes explained 80.94% of the overall variation in the studied set and separated male V. sylvestris with 3-colporate pollen grains on the right side from female individuals with inaperturate pollen grains on the left side of the 2-D plot. Two female individuals (with inaperturate pollen) were distinguished mainly by the equatorial axis length (Figure 3a). The leading phenotypic traits for this sample set were pollen length (P) and grain size-based gradient P/E having a strong but negative correlation with pollen width (E), together explaining 47.82% of the total variation along the first principal component. Exine ornamentation traits (perforation width and length) were strongly correlated to each other along the second principal component, but were independent from pollen dimensions (P, E and P/E) due to a very weak correlation. The number of exine perforations, despite significant differences observed among the genotypes, showed the weakest correlation to other traits (shortest vector). The longer the arrow, the stronger the correlation. The polar axis (P) and P/E ratio have strong positive correlation opposite to equatorial axis (E) length, with strong negative correlation toward P and P/E. Grouping of male individuals is mainly explained by P and P/E, and female individuals were negatively correlated with male individuals, being differentiated mainly by pollen width (E).
Ward's clustering method grouped individuals into two main subclusters ( Figure 4). This hierarchical clustering based on morphological traits (colpi values excluded) showed no particular grouping regarding the flower type or geographical origin of the individuals. The longer the arrow, the stronger the correlation. The polar axis (P) and P/E ratio have strong positive correlation opposite to equatorial axis (E) length, with strong negative correlation toward P and P/E. Grouping of male individuals is mainly explained by P and P/E, and female individuals were negatively correlated with male individuals, being differentiated mainly by pollen width (E).
Ward's clustering method grouped individuals into two main subclusters ( Figure 4). This hierarchical clustering based on morphological traits (colpi values excluded) showed no particular grouping regarding the flower type or geographical origin of the individuals.

Figure 4.
Ward's dendrogram of V. sylvestris genotypes constructed based on the Euclidean distance matrix from six studied morphological characteristics of pollen grains: polar axis (P), equatorial axis (E), P/E ratio, number of exine perforations, perforation width and perforation length, in Statistica.

Pollen Dimorphism
The microstructure of pollen from V. sylvestris in Croatia was revealed by scanning electron microscopy. Ten individuals from two populations were considered. The population in National Park Paklenica has been strictly protected for over 70 years; it belongs to the mountainous region with a shady forest habitat, with plenty of supporter plants for V. sylvestris. The population in Imotski is located further south, near the urban centers and with no strict protection of the site, fewer V. sylvestris individuals, poor supporter flora and soil characteristics, with dynamic water oscillations during the year [32]. Despite the population differences, pollen morphology and shape was not associated with population origin (Figures 3 and 4), though for the 3-colporate pollen of V. sylvestris from Turkey, although similar in morphology, a slight difference was observed in the shape at various localities [29]. The results for Croatian V. sylvestris are in accordance with findings from the Danube River area that showed a similarity in pollen morphological characteristics observed under light microscopy-a low variation in pollen size (P × E) despite long distances between populations [26]. An obvious difference was recorded in pollen between male and female V. sylvestris. Previous results on V. sylvestris revealed greater diversity among V. sylvestris itself rather than between V. sylvestris and V. vinifera [12]. This sample set, although small-sized regarding the number of individuals, encompassed more male than female individuals, reflecting male-female representations in natural habitats previously observed in Croatia and elsewhere [33][34][35]. The male predominance might correspond to the abundant pollen production and fertility of male pollen in wild populations [36].
All the pollen grains in this study, regardless of flower type, were radially symmetrical and isopolar, which is consistent with the pollen morphology of V. sylvestris studied in Turkey [29]. The exine ornamentation for all grains in this study was foveolate-perforate. Inceoglu et al. [29] reported on foveolate-rugulate and distinctly reticulate exine ornamentation at different pollen sites under SEM. Slight variations in ornamentation were observed at different sites of the same pollen grain or due to different sample preparation and the observation method used, as seen in studies [8,29]. The exine perforation observed under SEM in this study is in line with [29 cit. Faegri and Iversen 1964], as they reported that the exine sculpturing of Vitis was reticulate and foveolate-perforate under light microscopy. The entire exine sculpturing during pollen development might be affected by the expression of the INP1 gene [23].
The pollen grains of the male V. sylvestris in this study were tri-colporated and prolate in shape. This is usually observed in Vitis pollen, both in wild male and hermaphroditic cultivars [8,37]. Three-colporate pollen is linked to hermaphroditic flowers [20]. The prolate shape of pollen grains, with subtle variations, was observed in other V. sylvestris material [8,12,29]. Recently, prolate pollen was more frequently found in archaeological pollen than in modern pollen grains in Italy [8].
The pollen grains of the female V. sylvestris in this study were more oval (spheroidal to subprolate) than the pollen grains of the male individuals, and inaperturate. Quantitative traits such as pollen length (P), P/E ratio, number of perforations and colpus length were most informative for this sample set. Pollen length (P) varied significantly between the male-male and male-female individuals. Pollen width (E) did not vary significantly in this study, except for successfully differentiating males from females in accordance with their substantially different pollen shapes (Figures 2 and 3).
The pollen grains in this study were slightly larger in dimension than the ones observed in Western European V. sylvestris and in Italy [8,12]. This might be due to differences in pollen measurement methods- [8] used a light microscope, whereas the SEM method was used in this study-or possibly due to differences in sample sizes [38]. The pollen grains of female individuals in a previous study were found to be slightly thicker than the pollen grains of male individuals [8]. The pollen in Vitis is heavier compared to some species that have a light pollen grain to enable wind pollination. Though V. vinifera mostly depends on wind pollination, in V. sylvestris, pollination by insects seems to be even more important [2]. The inaperturate pollen of female individuals, with an excess deposition of sporopollenin in their oblate and a closed structure lacking germination spores, is associated with being a reward for pollinators [8], and could possibly explain the low pollen dispersal efficiency in Vitis.
The APT3 marker used in this study to genetically verify flower dimorphism in V. sylvestris clearly separated female from male individuals and confirmed the flower type determined by visual evaluation. Female individuals had typical female structures and were homozygous at the APT3 marker (266 allele) ( Figure 5) according to [31]. Male individuals all had typical male flower phenotypes and the allelic combination 266-397-466 at the APT3 marker, except for one individual (Pak13) that had a 266-466 allelic pattering. With the APT3 marker, it was possible to differentiate female individuals, but it was not possible to reliably distinguish males from hermaphrodites, which are occasionally found in natural habitats. Therefore, the APT3 marker was complementary to phenotypic observation for reliable identification of the flower type for V. sylvestris.
The multivariate analyses used in this study were both informative, especially PCA, which clearly differentiated male from female individuals based on morphological pollen data and revealed the phenotypic traits that contributed to pollen differentiation between male and female V. sylvestris.

Pollen Apertures
A female flower with a normally developed pistil and stamens, but short, curved or reflexed filaments, is related to the production of inaperturate pollen that seems to be sterile Plants 2022, 11, 1479 9 of 14 due to a lack of apertures preventing pollen germination [8]. Inaperturate pollen might also explain the poor fruit-set in V. sylvestris [8]. It is unknown whether the absence of pollen apertures in female individuals is sufficient to cause sterility in the grapevine [27]. Gallardo et al. [12] reported that inaperturate pollen in V. sylvestris is viable but lacking germination, and cultivars produce less fruit [12,20] or are sterile [39]. A recent study on the female cultivar 'Malbo Gentile' showed <5% of pollen with apertures, and inaperturate pollen, despite high viability (>90%), was unable to germinate. An in vitro germination test excluded the possibility that this cultivar produces cryptoaperturate pollen (pollen with an endoaperture that is not apparent in the surface view), indicating that pollen sterility is related to a physical barrier for pollen tube germination [8,40].
Female cultivars with inaperturate pollen are common, particularly among proles orientalis [41]. The cultivation of female cultivars is continually declining in modern vineyards due to fruit-set issues. Inaperturate pollen in cultivars is usually larger than in V. sylvestris individuals [12], similar to other results [8]. In a recent archaeological study, this type of pollen was proposed as a putative indicator of female Vitis sylvestris [8]. Interestingly, inaperturate pollen found in some species is correlated with an aquatic and moist, or nutrient-poor, habitat [40].
Other than the absence of apertures and pollen morphology itself, sterility in V. sylvestris might be related to poor nutrition uptake, e.g., zinc and boron at natural sites that are directly involved in the flower specification and fruit-set [21,39,42].
Alva et al. [39] reported the presence of various morphotypes of abnormal pollen in cultivars, all of which failed to germinate. The germination of abnormal pollen grains when supplemented with boron had only a minor effect compared to the treatment in cultivar with normal pollen, pointing to functional differences between normal and abnormal pollen. An excess of sporopollenin and exine deposition could result in infertile pollen [43]. A large deposition of these elements possibly leads to small and oval pollen grains lacking pores [12]. According to [30], two female individuals from the same population in this study differed significantly in the number of exine perforations and in pollen shape, which might be related to the inaperturate pollen specificities described.
The pollen grains of the male individuals in this study differed mostly in P/E ratio, number and length of perforations and colpus length. In Arabidopsis, a shorter aperture length proved to be related to lower transcript levels of the INP1 gene. Although INP1 is essential for aperture development, it is not the main factor defining aperture number, positions and morphology [24].

Sampling Locations and Plant Material
The studies were carried out with the pollen of V. sylvestris collected in 2018 from 10 individuals (eight male and two female), from two populations that were~175 km apart (Table 2, Figure 6). The population of National Park Paklenica, located in the mountainous region of Velebit in northern Dalmatia, has been strictly protected since 1949, with no or minimum human impact. Paklenica belongs to humid V. sylvestris forest habitats in Croatia, with vertically growing V. sylvestris lianas in the shade along both sides of the Paklenica stream and rich supporter-plant flora. The southern population in Imotski, in central Dalmatia, by contrast, is in the vicinity of urban centers. V. sylvestris grows horizontally on the rocky shores of the lake, receiving more sunlight during vegetation, with poor supporter vegetation, shallow, erosion-prone soil and pronounced yearly lake water oscillations, from complete coverage of individuals in water to a completely dry habitat [32]. Sampling was done during full bloom (in June) by excising V. sylvestris inflorescence and placing it in a clean and dry 50 mL polypropylene tube with a desiccator containing silica gel for each genotype, and storage at −20 • C.

Scanning Electron Microscopy of Pollen
Samples were transferred to the laboratory of the Faculty of Agriculture (Belgrade-Zemun, Serbia) for further pollen analyses. The analyses in this study were conducted according to [44][45][46]. Anthers were isolated from the flowers, collected in small vials and stored at 3-5 °C in a silica gel desiccator until analysis. After dehydration, the pollen was mounted for analysis using a fine brush on aluminum stubs covered with double-sided transparent tape and coated with a 0.02 µm-thick gold layer in a sputter coater BAL-TEC SCD 005 (Capovani Brothers Inc., Scotia, NY, USA). Observation was carried out at 15 kV with a scanning electron microscope (SEM) JEOL JSM-6390LV (Tokyo, Japan) at a magnification of 3000× (whole grain) and 15,000× (exine pattern). At least 30 randomly collected pollen grains per genotype were analyzed and presented as measurements of three replicates per genotype (10 pollen grains per replicate). The traits characterized were pollen size and shape (length, width and length/width ratio), length and width of the colpus, width of the mesocolpium and exine characteristics (the number of perforations per 25 µm 2 of the exine surface, perforation length and width), polarity, symmetry, aperture and exine ornamentation. Pollen shape characteristics were classified according to [30].

DNA Extraction and Marker Analysis
The selective DNA marker APT3 Indel [31] was employed to confirm the phenotype of the flower type and to differentiate female from male (or hermaphrodite) individuals.
Total genomic DNA was extracted from the young leaves using the NucleoSpin Plant II kit (Macherey-Nagel, Düren, Germany), with a working DNA concentration of 1 ng/µL.

Scanning Electron Microscopy of Pollen
Samples were transferred to the laboratory of the Faculty of Agriculture (Belgrade-Zemun, Serbia) for further pollen analyses. The analyses in this study were conducted according to [44][45][46]. Anthers were isolated from the flowers, collected in small vials and stored at 3-5 • C in a silica gel desiccator until analysis. After dehydration, the pollen was mounted for analysis using a fine brush on aluminum stubs covered with doublesided transparent tape and coated with a 0.02 µm-thick gold layer in a sputter coater BAL-TEC SCD 005 (Capovani Brothers Inc., Scotia, NY, USA). Observation was carried out at 15 kV with a scanning electron microscope (SEM) JEOL JSM-6390LV (Tokyo, Japan) at a magnification of 3000× (whole grain) and 15,000× (exine pattern). At least 30 randomly collected pollen grains per genotype were analyzed and presented as measurements of three replicates per genotype (10 pollen grains per replicate). The traits characterized were pollen size and shape (length, width and length/width ratio), length and width of the colpus, width of the mesocolpium and exine characteristics (the number of perforations per 25 µm 2 of the exine surface, perforation length and width), polarity, symmetry, aperture and exine ornamentation. Pollen shape characteristics were classified according to [30].

DNA Extraction and Marker Analysis
The selective DNA marker APT3 Indel [31] was employed to confirm the phenotype of the flower type and to differentiate female from male (or hermaphrodite) individuals.
Total genomic DNA was extracted from the young leaves using the NucleoSpin Plant II kit (Macherey-Nagel, Düren, Germany), with a working DNA concentration of 1 ng/µL. Reaction mixtures of 5 µL were prepared using KAPA Fast Multiplex PCR Kit (2x) (Kapa Biosystems, Wilmington, MA), 100 pmol of fluorescently labeled (FAM) reverse primer and~1 ng template DNA. PCR amplification was set up to three minutes initial denaturation at 95 • C, 30 cycles of denaturation at 95 • C (15 s), annealing at 60 • C (30 s), extension at 72 • C (30 s) and final extension for seven minutes at 72 • C. PCR amplificates were analyzed using capillary electrophoresis on an ABI 3730xl Genetic Analyzer (Applied Biosystems, Waltham, MA, USA) with GeneScan-LIZ 500 standard. The peaks were identified with GeneMapper 4.0 software (Applied Biosystems, Waltham, MA, USA).

Statistical Analysis
The mean value for pollen grains was reported with standard deviations. Data were statistically analyzed using a one-way analysis of variance (ANOVA). The significance of the differences between the mean values was determined using Tukey's test and considered significant at p ≤ 0.05. Genotypes were clustered into similarity groups using Principal Component Analysis (PCA) and Ward clustering methods based on tested characteristics, except for colpi values (colpus length, width and mesocolpium width) that are lacking in female individuals. All statistical analyses were performed using the software package STATISTICA v. 14.0 (TIBCO Software Inc, Palo Alto, CA, USA).

Conclusions
Scanning electron microscopy revealed the structure of the pollen grains of male and female V. sylvestris in Croatia for the first time. The analyses clearly differentiated male from female pollen morphotypes, but no separation by population was observed. All pollen grains of V. sylvestris were isopolar, radially symmetrical and foveolate-perforate ornamented, which is in line with previous, similar studies. Two pollen morphotypes were observed: 3-colporate pollen with a prolate shape recorded in male individuals and inaperturate pollen with a spheroidal to subprolate shape in two female individuals that significantly differed in the number of exine perforations. Quantitative pollen dimensions described by polar axis, P/E ratio, number of perforations and colpus length were most polymorphic for the analyzed V. sylvestris set. PCA and cluster analysis was informative with regard to pollen sex dimorphism. This study opens the door for further studies on Vitis flower morphology and sterility issues by testing pollen germination and comparing the INP1 gene between hermaphroditic and female genotypes of V. sylvestris.