Morphometric Analysis of Grape Seeds: Looking for the Origin of Spanish Cultivars

Abstract

The Vitis IMIDRA collection contains 3699 entries, representing a significant percentage of the variation in traditional and commercial Vitis cultivars used in Spain. The classification and identification of new entries are currently conducted based on ampelography and molecular methods. Here, we propose a new method of classification of the cultivars based on seed morphology and its application to a total of 224 varieties from the collection. Based on seed shape, fourteen groups have been defined according to the similarity of the seeds, with geometric figures used as models. The new models are Cariñena Blanca, Chardonnay, Parraleta, and Parduca, defining new groups to be added to the ten groups previously described. The study results in 14 groups comprising the Spanish cultivar’s seed shape and morphological variation. Seed morphology can help to identify varieties cultivated in the past through archaeological finds.

1. Introduction

The European grapevine (Vitis vinifera L., Vitaceae) comprises more than 10,000 varieties widely cultivated throughout the world. Proof of this is the 7.4 million hectares of vineyards planted for winemaking, table grapes, sultanas, or juice concentrate, among others [1]. Currently, around 63% of this area belongs to Europe [2].

The Quaternary glaciations determined the distribution of the genus Vitis, resulting in new species [3]. The populations taking refuge in the Transcaucasian area and the Mediterranean basin gave rise to the current Vitis vinifera L., or European vine [4], recorded in the Iberian Peninsula from the Lower and Middle Pleistocene, some 780,000 years ago in Atapuerca (Burgos, Spain) [5] and El Padul (Granada, Spain) [6]. Wild or cultivated seeds can be found in burials as gifts, together with grape juice or wine [7,8].

The Eurasian vine comprises two subspecies: Vitis vinifera L. subsp. sylvestris (Gmelin) Hegi, dioecious with sexual dimorphism, and Vitis vinifera L. subsp. sativa (DC.) Hegil, whose plants are, in general, hermaphrodite [9,10]. Exceptionally, plants are found with functionally female flowers, as in the table variety Ohanes [11]. They are distributed in natural ecosystems from Portugal to the Indus Kush massif, as well as in the Maghreb [12]. Cultivated grapevine originates from selection from mutations that resulted in hermaphrodite cultivars [13]. The distribution comes from the centers of domestication in regions in Mesopotamia and the Middle East [14].

Wild female vines produce berries with rounded seeds that are much smaller in size and easier to germinate than those of cultivated varieties [15]. The morphometric differences are useful for distinguishing between wild and cultivated seeds found in archaeological sites [15], as well as for inferring their degree of domestication [16]. The sub-globose shape of wild seeds [17] is supported by morphometric parameters from different studies comparing the seeds of cultivated and wild plants using the Stummer Index (SI) or other parameters [18,19].

The genetic analysis of vine plant material shows the relationship between varieties as well as their origin. DNA analysis for the identification of chlorotypes of autochthonous varieties cultivated in the Iberian Peninsula shows that 70% contain chlorotype A, as well as the wild populations in the area. In the Transcaucasian area, the proportion of the dominant chlorotype in wild and cultivated varieties corresponds mainly to types C and D [20]. It shows the influence of the varieties introduced in the Iberian Peninsula by Arabs, Phoenicians, Greeks, Romans, and Carthaginians, along with the hybrids resulting from the crosses with local varieties, giving rise to today’s vast genetic heritage [21]. Spain boasts a heritage of 530 indigenous varieties, with many of them currently in minority cultivation. Identifying the varieties cultivated in antiquity and understanding their relationship with those cultivated today pose complex challenges. The book by Alonso de Herrera contains the first mention and description in detail for a good number of varieties, some of which are still cultivated today, such as ‘Cigüente’, ‘Jaén’, ‘Hebén’, ‘Alarije’, ‘Palomina’, ‘Castellano Blanco’, ‘Tortozón’, ‘Albillo’, ‘Torrontés’ and what is now known to be ‘Tempranillo’ [22].

Andalusian authors propagated the vines by mass sowing of seeds, a fact that supports the fact that the ‘Hebén’ or ‘Gibi’ variety is the mother of many varieties in the Iberian Peninsula [11]. The development of ampelography during the 20th century and later molecular techniques have made it possible to advance in the identification of vine plant material and its genetic relationships. However, there is still important information missing, especially about the geographical and temporal origin of the known varietal heritage.

The development of image analysis protocols and software [23] has allowed the identification of charred grape seeds found in Bronze Age archaeological sites in Sardinia. Archaeo-botanical surveys in Spain and the rest of Europe continue to find evidence of the presence of vines and their use in archaeological contexts [24], and the use of radiocarbon dating has clarified the age of the seeds found. The development of morphometric models [25] for ancient vine seeds could help to clarify the origin, development, and domestication of Vitis specimens [26,27].

In addition to artificial vision protocols and Fourier transform analysis, an alternative method to describe and classify seed shapes consists of the comparison of the seeds with geometric figures. In this method, the J index is the percentage of area shared between seed images and geometric figures [28]. It has been applied for the description of seed shape in the model plant Arabidopsis thaliana (L.) Heynh (Brassicaceae) [29], as well as in the model legumes Lotus japonicus (Regel) K.Larsen and Medicago truncatula Gaertn. (Fabaceae) [30]. Shape quantification by the J index has been applied in species of a diversity of families with different geometric models [31,32,33,34], including the Vitaceae Juss [35,36].

In a recent study with 38 representative cultivars of the IMIDRA collection, ten groups were proposed according to the similarity of their seeds with geometric figures [36]. The groups were given the names of their representative cultivars: ‘Listán Prieto’, ‘Silvestris’, ‘Albillo Real’, ‘Hebén’, ‘Moscatel’, ‘Tortozón’, ‘Teta de Vaca’, ‘Doña Blanca’, ‘Airén’ and ‘de Cuerno’. The groups were ordered in this list by their increasing aspect ratio (length/width), with each of them containing the cultivars whose seeds showed a high percentage of similarity to a corresponding geometric model. Two different types were recently described according to their apex curvature. The seeds of groups Listán Prieto, Albillo Real, Moscatel, and Doña Blanca have, in general, their apex plane at the top, while seeds in the groups Hebén, Tortozón, and de Cuerno have acute peduncles [37].

The aim of this work is to compare the seed morphology of grapevine cultivars in Spain with a total of fourteen geometric models obtained from the algebraic modification of Models 7 and 8 [35,36] and to confirm the genetic relationships of these varieties by elucidating their geographical and temporal origin. The varieties studied have been selected for their historical relevance, geographical distribution, and importance in Spanish viticulture. The seed models generated in this work could also be of great help in archaeological discoveries related to wine-related cultures.

2. Materials and Methods

2.1. Plant Material

Characteristics distinctive of the cultivars analyzed are presented in

Table 1. List of the 224 varieties studied. All seeds were collected in the vineyards of the El Encin estate in Alcalá de Henares, Madrid, where the cultivars are maintained. Seeds were extracted from berries and kept under dry conditions for measurements. BA: Badajoz; CC: Cáceres; CO: Córdoba; FR: France; H: Huelva; J: Jaen; NA: Navarra; S: Santander; SE: Sevilla.

Variety Name
Airén	Castellana Blanca	H 6.1 *	Moravia Dulce	SE 2.4 *
Alarije	Cayetana Blanca	H 6.5 *	Morenillo	SE 2.6 *
Albana	CC 1.5 *	Hebén	Morisca	SE 2.7 *
Albarín Blanco	CO 4.7 *	Imperial	Moristel	S 3.5b *
Albillo Real	Corchera	Italia	Moscatel Alejandría	Sabaté
Albariño	Crepa	J 1.4 *	Moscatel Angües	Sabro
Albillo de Granada	Cuatendrá	J 2.4 *	Moscatel G. Menudo	Salvador
Albillo del Pozo	Chardonnay	Jaén Rosado	Moscatel Rosado	Sanguina
Albillo Mayor	Chasselas Blanc	Jaén Tinto	NA 2.4b *	Sauvignon Blanc
Albillo Real	de Cuerno	Jarrosuelto	NA 3.2 *	Señá
Alcañón	Derechero	Jerónimo	Negra Rayada	Sousón
Aledo	Diega	Juan García	Negreda	Sumoll
Aledo Real	Diega 2	Lado	Ohanes	Syrah
Arcos	Dominga	Legiruela	Ondarrabi Beltza	Tempranillo
Aúrea	Doña Blanca	Listán del Condado	Ondarrabi Zuri	Terriza
Azargón	Doradilla	Listán Negro	Palomino Fino	Teta de Vaca
BA 2.3 *	Epero de Gall	Listán Prieto	Palote	Tinto Bastardo
Bastardo Blanco	Espadeiro	Loureira	Pampolat de Sagunto	Tinto Fragoso
Bastardo Negro	Estaladiña	Lucomol	Pardillo	Tinto Jeromo
Batista	Excursach	Malbec	Parduca	Tinto Navalcarneo
Beba	FR 1.1 *	Malvar	Parellada	Tinto Pámpana Blanca
Benedicto	FR 1.4 *	Malvasia Aromática	Parraleta	Tinto Velasco
Benedicto falso	Fernao Pires	Macabeo	Pedrol	Torralba
Bermejuela	Ferral	Mandón	Pedro Ximenez	Torrontés
Blanquiliña	Ferrón	Mandregue	Perruno	Tortosí
Bobal	Fogoneu	Manses de Capdell	Picapoll BlancoPicapoll Tinto	Tortozón
Brancellao	Folle Blanche	Manses de Tibbus	Pinot Noir	Tortozona Tinta
CA 13.3 *	Forastera	Manto Negro	Pintada	Treixadura
CA 13.4 *	Forcallat Tinta	Mantuo de Pilas	Planta Fina	Trepat
CA 13.6 *	Gabriela	Marfal	Planta Mula	Trobat
CA 2.9b *	Garnacha Blanca	Marfileña	Planta Nova	Trincadeira das Pratas
Cabernet Franc	Garnacha Peluda	Maturana Blanca	Prieto Picudo	Verdejo
Cabernet Sauvignon	Garnacha Roja	Mazuela	Puerto Alto	Verdejo de Salamanca
Cadrete	Garnacha Tinta	Mencía	Quigat	Verdil
Cagarrizo	Garrido Fino	Merenzao	Ragol	Vidadillo
Caiño Blanco	Garrido Macho	Merlot	Ratiño	Vijariego Blanco
Caiño Longo	Gewürztraminer	Merseguera	Rayada Melonera	Viognier
Caiño Tinto	Giro Negro	Miguel de Arco	Riesling	Viura
Callet	Godello	Mollar Cano	Rocia	Xarello
Cañorroyo	Gonfaus	Monastrell	Rojal Tinta	Xarello Rosado
Cardinal	Gorgollasa	Mondragón	Rosseti	Zalema
Cariñena Blanca	Graciano	Montonera	Rubeliza	Zurieles
Cariñena Roja	Granadera	Montúa	Rufete	CA 2.9 *
Carrasquín	Grand Noir	Morate	Rufete Serrano
Castañal	Gualarido	Moravia Agria	SE 2.1 *

* sylvestris accession.

.

2.2. Photography

Seed images were taken with a Nikon D80 10.2 megapixel camera (Nikon, Tokyo, Japan).

2.3. Similarity between Seeds and the Models: J Index

The J index is calculated by a comparison between two images: the seed image and the geometric model. Figures of the model (.png) are superposed to images of 20 seeds in each cultivar, searching for maximum similarity. Adobe Photoshop archives (PSD; Photoshop CC 2020; Adobe, San José, CA, USA) are kept for each cultivar, containing twenty seeds and the corresponding models superimposed. From each of these images, two new archives are kept: one with the model in black and the other with the model in white. The Image J program [38] gives, respectively, for each of these images the values of total area (T, being the model silhouette in black, all the area is considered) and area shared between the model and the seeds (S, being the model silhouette in white, the area measured is limited to the shared area between seed, and the model). For each cultivar with a given model, the J index is the mean ratio of the S/T value obtained for 20 seeds (Figure 1).

2.4. Seed Shape in Cultivars: The Average Silhouette

Average silhouettes were obtained as described by Cervantes et al. [36]. The comparison of each of the average silhouettes obtained for the cultivars with the ten models gave a first idea of the model that best fits each cultivar. The J index was quantified with those cultivars resembling more the average silhouette, and the cultivar was attributed to the group where the J index with the corresponding model was superior to 90. In cases where a J index value superior to 90 was obtained with more than one model, then the cultivar is attributed to the group corresponding to the model with which it gave a higher J index score. Cultivars with high variation in shape were attributed to a group with J index scores below 90. New models (and new groups) were defined for those cultivars whose seeds could not be attributed to the existing groups.

2.5. New Models

Four new models were obtained by the modification of equation 8 [36] by searching for geometric figures similar to the average silhouettes of cultivars whose seeds did not fit any existing model.

2.6. Multivariate Analysis

Multivariate analysis was performed in R Studio, V.1.2.1335 [39]. Raw data were analyzed to check if the dataset was homoscedastic, which means homogeneity of variance and a normal distribution. A normalization of raw data to lower the weight of outliers prior to starting with multivariate analysis was required and achieved using the powerTransform function. A lambda value to transform the data was used according to mathematical procedures. This function uses the maximum likelihood-like approach of Box–Cox [40] to select a transformation of a univariate or multivariate response for normality. Principal component analysis is a procedure for dimension reduction used to see the total variation of variables and individuals in multidimensional data. The distribution of the species is based on their value for each parameter combined. This offers the possibility to gather species among them according to specific traits or, like in this analysis, to specific geometrical model indexes. The hierarchical clustering from the PCA results was performed according to the Euclidean distance and applying the ward.D2 method to minimize the total within-cluster variance [41]. In order to identify the statistical differences, a multivariate analysis of variance (MANOVA), a procedure for comparing multivariate sample means using covariance, was applied. From the morphological description data, roundness was removed from the analysis to be redundant with the Stummer index.

3. Results

3.1. New Models

Four new models were obtained based on the seed shapes of the cultivars ‘Parraleta’, ‘Cariñena Blanca’, ‘Chardonnay’, and ‘Parduca’. The corresponding equations are as follows:

Model Parraleta (Model 11):

$$\left(\frac{25\sqrt{33-x^2}}{22}+\frac{1600}{264+17x^4}-\frac{9y}{10}\right)\left(\sqrt{33-x^2}-\frac{25}{33+55x^2}+\frac{9y}{10}\right)=0$$

(1)

Model Cariñena Blanca (Model 12):

$$\left(\sqrt{\frac{165-6x^2}{5}}+\frac{1800}{324+25{\left(12x^{18}\right)}^{1/5}}-\frac{9y}{10}\right)\left(\frac{50}{11(5+x^2+x^4)}-\frac{5}{22}\sqrt{825-30x^2}-\frac{9y}{10}\right)=0$$

(2)

Model Chardonnay (Model 13):

$$\left(y-\sqrt{33-x^2}-\frac{500}{115+x^6}\right)\left(\sqrt{33-x^2}-\frac{4}{5+x^2+x^4}+\frac{22}{25}y\right)=0$$

(3)

Model Parduca (Model 14):

$$\left(\frac{25\sqrt{33-x^2}}{22}-\frac{50}{11(5+x^2+x^4)}+y\right)\left(-\frac{27}{16}\sqrt{33-x^2}-\frac{150}{33+\frac{11x^{14}}{\mathrm{200,000}}}+y\right)=0$$

(4)

The new models complete the range of the Stummer index (SI) covered by the 10 models described before [36] (

Table 2. Values of the Stummer index, the J index with the best-fitting cultivar, the mean J index of all cultivars for each model (max/min), and the number of cultivars corresponding to the 14 models used in this work. Each of the groups is defined by a geometric model that shares the name of a representative cultivar in this group. The specificity of a given model increases with the difference between the J index value with the best-fitting cultivar and the mean J index across cultivars.

Model		Stummer Index	J Index (Best Fitting Cultivar)	J Index Mean in the Cultivars (max/min)	N
1	Listán Prieto	0.81	90.6	90.1 (90.6/89.6)	4
2	Chardonnay	0.78	90.8	90.7 (90.8/90.6)	2
3	Sylvestris	0.72	91.9	90.3 (91.9/89.6)	22
4	Hebén	0.70	91.4	90.5 (91.4/89.6)	22
5	Albillo Real	0.69	91.5	90.3 (91.5/89.6)	27
6	Tortozón	0.68	90.0	89.8 (90.0/89.6)	6
7	Moscatel	0.64	93.0	90.9 (93.0/89.7)	60
8	Teta de Vaca	0.62	91.7	90.6 (91.7/89.9)	21
9	Cariñena Blanca	0.60	91.1	90.5 (91.1/89.8)	3
10	Parduca	0.58	89.7	89.7 (89.7/89.6)	3
11	Airén	0.58	90.9	90.6 (90.9/90.2)	3
12	Doña Blanca	0.57	91.1	90.2 (91.1/89.6)	45
13	Parraleta	0.56	90.1	89.7 (90.1/89.6)	4
14	de Cuerno	0.45	89.3	89.3	1

). Thus, SI for Chardonnay is intermediate between those of Listán Prieto and Sylvestris, Cariñena Blanca and Parduca are intermediate between Teta de Vaca and Airén, and Parraleta between Doña Blanca and de Cuerno. In addition, the new models complete the series that were defined by different curvature values at the apex of the peduncle [37]. The series of seeds with an acute peduncle (a maximum peak of curvature) formed by groups Hebén and Tortozón is completed by Cariñena Blanca (Figure 2). While the series of cultivars whose seeds have a plane peduncle is formed by Listán Prieto, Albillo Real, Moscatel, and Doña Blanca (Figure 3), a new series of cultivars with a rounded peduncle is now represented by Chardonnay, Parduca, and de Cuerno (Figure 4). The models Moraté and Parraleta are included in the series formed by the groups with an intermediate-type peduncle (Figure 5) [37].

3.2. Distribution of 223 Cultivars in the 14 Groups Defined by the Models

The new models described here, together with those described before [36], represent a total of 14 groups that include 223 of the 224 cultivars under study. The only cultivar unattributed to any of the 14 groups is ‘Rayada Melonera’, whose special characteristics and diverse morphological seed types will be discussed elsewhere. The remaining 223 cultivars are classified into 14 groups, as indicated below (

Table 3. Distribution of the 223 cultivars under study in the 14 morphological groups defined by the models. The cultivars are grouped by the similar morphology of their seeds. J index values indicate the percent similarity with the corresponding model.

Model (Number of Cultivars)	Cultivars (J Index Values)
Listán Prieto (4)	Cardinal (89.9), Listán Prieto (90.7), Montonera (89.6), Tortozona Tinta (92.6)
Chardonnay (2)	Chardonnay (90.8), Trobat (90.6)
Sylvestris (22)	BA 2.3 (91.2), Benedicto (89.7), CA 13.3 (89.6), CA 13.4 (91.0), CA 13.6 (90.9), CC 1.5 (90.5), Diega (90.7), FR 1.1 (90.2), Gonfaus (90.2), H 6.1 (89.6), H 6.5 (89.7), J 1.4 (89.6), J 2.4 (90.2), Lucomol (91.0), Merlot (89.9), NA 2.4b (90.0), Negra Rayada (89.6), Rufete (89.8), S 3.5b (89.7), SE 2.1 (90.1), SE 2.7 (90.6), Zurieles (89.5)
Hebén (22)	Albana (91.4), Aledo (90.6), Aledo Real (90.8), Corchera (90.7), Chasselas Blanc (90.1), Ferrón (90.9), Granadera (90.7), Hebén (90,0), Macabeo (91.3), Malvar (90.3), Marfal (90.3), Miguel de Arco (89.6), Pardillo (90.2), Parellada (90.6), Planta Mula (90.3), Planta Nova (89.9), Rosetti (89.8), Señá (91.3), Terriza (90.4), Tinto Fragoso (90.1), Viura (90.4), Zalema (91.6)
Albillo Real (27)	Alarije (90.0), Albillo Real (90.3), Azargón (89.6), Bastardo Negro (89.8), Blanquiliña (91.1), Bruñal (92.2), Cabernet Sauvignon (91.5), Carrasquín (90.5), Cayetana Blanca (90.6), Crepa (90.3), Giro Negro (90.5), Graciano (90.3), Jerónimo (90.5), Juan García (90.4), Mantúo de pilas (90.0), Merenzao (90.6), Moravía dulce (89.6), Morenillo (90.7), Ondarrabi Zuri (89.7), Pinot Noir (90.3), Planta Fina (89.7), Ratiño (89.7), Rojal tinta (89.9), Tempranillo (90.3), Tinto bastardo (90.8), Torrontés (90.0), Treixadura (91.0)
Tortozón (6)	Garrido Fino (89.7), Imperial (88.9), Moraté (89.8), Picapoll Blanco (90.0), CA 2.9 (89.6), Tortozón (89.9)
Moscatel (60)	Albillo del Pozo (91.5), Albillo Mayor (90.8), Arcos (90.7), Áurea (91.7), Beba (91.1), Benedicto falso (90.2), Cagarrizo (90.8), Caíño Blanco (89.8), Caíño Longo (90.4), Caíño Tinto (91.2), Castellana Blanca (90.5), CO 4.7 (89.9), Cuatendrá (89.8), Diega 2 (91.7), Doradilla (93.0), Espadeiro (91.4), Fernâo Pires (91.0), Ferral (90.3), Forastera (91.0), Gabriela (92.0), Garnacha peluda (90.2), Garnacha roja (91.4), Garnacha Tínta (91.2), Gewürztraminer (92.0), Grand Noir (91.6), Jaén Rosado (90.1), Jaén Tinto (91.6), Listán del Condado (91.1), Malbec (90.3), Malvasía Aromática (90.7), Mansés de Capdell (91.2), Marfileña (90.4), Maturana Blanca (90.7), Merseguera (91.6), Mollar Cano (91.1), Morisca (90.6), Moristel (91.2), Moscatel de Alejandría (90.3), Moscatel de Angüés (91.1), Moscatel de Grano Menudo (91.4), Moscatel Rosa (92.1), Negreda, (91.2), Ondarrabi Beltza (90.5), Palomino Fino (90.9), Pampolat de Sagunto (91.3), Perruno (91.7), Picapoll Tinto (90.0), Prieto Picudo (90.9), Puerto Alto (90.3), Quigat (91.5), Ragol (89.7), Riesling (90.4), Rocía (92.1), Rubeliza (90.2), Salvador (91.4), Tinto Jeromo (90.8), Tinto de Navalcarnero (90.8), Tinto Velasco (91.4), Trepat (90.4), Verdil (90.3)
Teta de Vaca (21)	Alcañón (90.2), Cadrete (90.2), Cariñena Roja (89.9), Dominga (90.6), Esperó de Gall (89.9), Forcallat Tinta (90.0), Gorgollasa (91.7), Mandregue (90.8), Montúa (90.3), NA 3.2b (91.4), Ohanes (90.0), Palote (91.0), Pedrol (91.3), Rufete Serrano (90.8), Sanguina (90.9), SE 2.6 (91.1), Teta de Vaca (90.4), Tortosí (90.0), Verdejo (90.1), Verdejo de Salamanca (90.6), Xarel lo (90.1)
Cariñena Blanca (3)	Bermejuela (90.0), Cariñena Blanca (90.6), Moravía Agria (91.1)
Parduca (3)	Parduca (89.7), Vijariego Blanco (89.7), Viognier (89.6)
Airén (3)	Airén (90.4), Bobal (90.3), Mazuela (90.1)
Doña Blanca (45)	Albarín blanco (89.8), Albariño (89.6), Albillo de Granada (90.1), Bastardo Blanco (90.1), Batista (90.1), Brancellao (90.0), CA 2.9b (89.9), Cabernet Franc (89.9), Callet (90.4), Cañaroyo (90.8), Castañal (90.5), Derechero (91.1), Doña Blanca (91.0), Estaladiña (89.7), Excursach (89.9), Fogoneu (89.7), Folle Blanche (89.7), FR 1.4 (90.7), Garnacha blanca (89.7), Godello (90.2), Gualarido (90.1), Italia (Moscatel Romano) (89.9), Jarrosuelto (90.8), Lado (91.1), Legiruela (89.8), Listán negro (90.9), Loureira (90.6), Mandón (90.7), Mansés de Tibbus (89.8), Manto negro (90.4), Mencía (90.5), Monastrell (90.2), Pedro Ximénez (90.5), Pintada (90.0), Sabro (90.0), Sauvignon Blanc (89.6), SE 2.4 (89.7), Sousón (89.6), Sumoll (89.8), Syrah (89.6), Tinto de la Pámpana Blanca (90.8), Torralba (90.6), Trincadeira das Pratas-Allarén (89.6), Vidadillo (89.6), Xarel lo rosado (89.6)
Parraleta (4)	Garrido macho (90.1), Mondragón (89.6), Parraleta (89.6), Sabaté (89.6)
de Cuerno (1)	de Cuerno (88.8)

).

3.3. Variation in Size and Shape in the Cultivars

The range of variation in seed dimensions was compatible with the data reported by Rivera et al. [18] (

Table 4. Mean values of area, perimeter, length, width, Stummer index, circularity, and roundness. Minimum and maximum values are given for individual seeds and for the means of cultivars. Data between parentheses correspond to cultivars. In cases where there is no name indicated between parentheses, this means that the maximum and minimum values of the cultivar correspond to the same cultivar than the maximum or minimum values obtained for individual seeds. See the Supplementary Tables for details.

	Mean	Min	Max
Area	17.46	9.16 Sylvestris (‘H 6.5’), (11.79)	30.26 ‘de Cuerno’, (28.30)
Perimeter	17.51	11.59 Sylvestris (‘H 6.5’), (13.49)	26.80 ‘de Cuerno’, (25.44)
Length	5.91	3.74 Sylvestris (‘H 6.5’), (4.53)	9.75 ‘de Cuerno’, (9.28)
Width	3.75	2.77 ‘Sabaté’, (3.02 ‘Folle Blanche’)	5.29 ‘Caiño Tinto’, (4.59 ‘Juan García’)
Stummer Index	0.64	0.37 ‘de Cuerno’, (0.42)	0.88 ‘Cardinal’, (0.80 ‘Listán Prieto’)
Circularity	0.71	0.37 ‘Riesling’, (0.55, ‘de Cuerno’)	0.86 Sylvestris (‘H 6.5’), (0,73)
Roundness	0.64	0.38 ‘de Cuerno’, (0.42)	0.88 ‘Cardinal’, (0.80 ‘Listán Prieto’)

). Seed length was comprised between 3.74 and 9.75 mm (3.73 and 8.10 in [18]); width was between 2.77 and 5.29 (1.79 and 5.23 in [18]). The mean values of the Stummer index were between 0.42 and 0.80 (0.43 and 1.06 in [18]), with maximum scores in the cultivar ‘Listán Prieto’ and minimum scores in ‘de Cuerno’. The minimum values of area, perimeter, and length were obtained in the ‘Sylvestris’ varieties, while maximum values of these magnitudes were obtained in cultivar ‘de Cuerno’.

3.4. Morphological Groups by Series

The groups were divided into four series according to the shape at the end of the seed peduncle. The peduncle may end in an acute, plane, rounded, or intermediary-indeterminate shape. Types of peduncle acute and plane were defined by their curvature values [37].

Tables in the Supplementary Data contain the mean and standard deviation, minimum and maximum values for area, Stummer index, and J index for a total of 223 cultivars grouped into 14 groups. Series 1 contains the three groups that include seeds with acute peduncles: Hebén, Tortozón, and Cariñena Blanca. Series 2 contains four groups of seeds with a plane peduncle: Listán Prieto, Albillo Real, Moscatel, and Doña Blanca. Series 3 contains three groups of round-ended peduncles: Chardonnay, Parduca, and de Cuerno. Finally, Series 4 contains four groups of intermediate or undefined peduncle types (Sylvestris, Teta de Vaca, Airén, and Parraleta).

3.4.1. Groups with Curvature Type 1 (Acute Peduncle)

This section includes the seeds of 26 cultivars in three groups, according to the models drawn in Figure 2: Hebén (22), Tortozón (5), and Cariñena Blanca (1), and a summary of the results with the groups of Curvature Type 1 (

Table 5. Summary of results for the groups with Curvature Type 1. The values given correspond to minimum and maximum values among the means obtained for all the cultivars in each group (for details, see Supplementary Tables S1–S3).

	Area	Stummer Index	J Index
Hebén (22 cultivars)	14.9–22.4	0.64–0.75	89.6–91.4
Tortozón (6 cultivars)	16.7–20.3	0.61–0.64	89.6–90.0
Cariñena Blanca (3 cultivars)	15.3–17	0.58–0.60	89.8–90.6

).

Supplementary Table S1 contains mean values and standard deviations, as well as minimum and maximum values for the area, Stummer index, and J index of the twenty-two cultivars comprised in group Hebén. The mean area values in the cultivars belonging to group Hebén comprise 14.9 and 22.4 mm². The mean Stummer index values comprise 0.64 and 0.75. The mean J index values comprise 89.6 and 91.4.

Figure 6 contains images of the model Hebén, the superimposed silhouettes of 30 seeds of cultivars ‘Aledo’, ‘Macabeo’, and ‘Malvar’, the average silhouette for these three cultivars obtained by the pixel profile of the image with the silhouettes, and four representative seeds of each of the three cultivars (Upper row: ‘Aledo’; Middle row: ‘Macabeo’; Lower row: ‘Malvar’).

Supplementary Table S2 contains mean data for the area, Stummer index, and J index of the six cultivars comprised in group Tortozón. Mean area values comprise between 16.7 and 20.3 mm². The mean Stummer index values comprise 0.61 and 0.64. The mean J index values comprise 89.6 and 90.0.

Figure 7 contains images of the model Tortozón, the silhouettes of 30 seeds of cultivar Imperial, the average silhouette for this cultivar obtained by the pixel profile of the image with the silhouettes, and four representative seeds of cultivar Imperial.

Supplementary Table S3 contains mean data for the area, Stummer index, and J index of the three cultivars comprising the group Cariñena Blanca. Mean area values range between 15.3 and 17 mm². The mean Stummer index values range between 0.58 and 0.60. Mean J index values range between 89.8 and 90.6.

Figure 8 contains an image of the model Cariñena Blanca, an image with the silhouettes of 30 seeds of cultivar Cariñena Blanca, the average silhouette for this cultivar obtained by the pixel profile of the image with the silhouettes, and four representative seeds of cultivar ‘Cariñena Blanca’.

3.4.2. Groups with Curvature Type 2 (Peduncle Plane at the Top)

This series includes data for seeds of 138 cultivars in four groups: Listán Prieto (4), Albillo Real (25), Moscatel (62), and Doña Blanca (47)

Supplementary Table S4 contains mean data for the area, Stummer index, and J index of the four cultivars in group Listán Prieto. The mean area values in this group range between 15.1 and 19.1 mm². The mean Stummer index values range between 0.75 and 0.77. Mean J index values range between 89.6 and 90.6.

Figure 9 contains an image of the model Listán Prieto, the silhouettes of 30 seeds of cultivars ‘Cardinal’ and ‘Tortozona Tinta’, the average silhouette for these cultivars obtained by the pixel profile of the image with the silhouettes, and four representative seeds of each of the two cultivars (Upper row: ‘Cardinal’; Lower row: ‘Tortozona Tinta’).

Supplementary Table S5 contains mean data for the area, Stummer index, and J index of the 25 cultivars comprised in group Albillo Real. The mean area values of the cultivars belonging to the group Albillo Real range between 14.2 and 23.1 mm². The mean Stummer index values range between 0.64 and 0.73. Mean J index values range between 89.6 and 91.5.

Figure 10 contains an image of the model Albillo Real, the silhouettes of 30 seeds of cultivars Graciano y Tempranillo, the average silhouette for these cultivars obtained by the pixel profile of the image with the silhouettes, and four representative seeds of each of the two cultivars.

Supplementary Table S6 contains mean data for the area, Stummer index, and J index of the 62 cultivars included in group Moscatel. The mean area values in the cultivars belonging to group Moscatel range between 12.1 (‘Caíño longo’) and 24.9 mm² (‘Negreda’). The mean Stummer index values range between 0.58 and 0.69. Mean J index values range between 89.7 and 93.0.

Figure 11 contains an image of the model Moscatel, the silhouettes of 30 seeds, the average silhouette obtained by the pixel profile of the image with the silhouettes, and four representative seeds of ‘Moscatel de Grano Menudo’, ‘Garnacha Tinta’, ‘Gewurztraminer’ and ‘Jaén Tinto’.

Supplementary Table S7 contains mean data for the area, Stummer index, and J index of the 47 cultivars comprised in group Doña Blanca. The mean area values of the cultivars belonging to the group Doña Blanca range between 12.9 and 22.9 mm². The mean Stummer index values range between 0.54 and 0.69. The mean J index values range between 89.6 and 91.1.

Figure 12 contains an image of the model Doña Blanca, the silhouettes of 30 seeds, the average silhouette obtained by the pixel profile of the image with the silhouettes, and four representative seeds of ‘Albariño’, ‘Bastardo Blanco’, ‘Mencía’ and ‘Pedro Ximenez’.

Table 6. Summary of results for the groups with Curvature Type 2. The values given correspond to minimum and maximum values among the means obtained for all the cultivars in each group (for details, see Supplementary Tables S4–S7).

	Area	Stummer Index	J Index
Listán Prieto	15.1–19.1	0.75–0.77	89.6–90.6
Albillo Real	14.2–23.1	0.64–0.73	89.6–91.5
Moscatel	13.2–24.9	0.58–0.69	89.7–93.0
Doña Blanca	12.9–22.9	0.54–0.69	89.6–91.1

contains a summary of the results of the groups with Curvature Type 2.

3.4.3. Groups with a Rounded Peduncle

This series included six cultivars in three groups: Chardonnay (2 cultivars), Parduca (3 cultivars), and de Cuerno. Supplementary Table S8 contains the mean data for the area, Stummer index, and J index of the two cultivars comprising the group Chardonnay. The mean area values in this group range between 16.0 and 16.9 mm². The mean Stummer index values are 0.74. The mean J index values range between 90.6 and 90.8.

Figure 13 contains an image of the model Chardonnay, an image with the silhouettes of 30 seeds of this cultivar, the average silhouette obtained by the pixel profile of the image with the silhouettes, and four representative seeds of the cultivar Chardonnay.

Supplementary Table S9 contains mean data for the area, Stummer index, and J index of the cultivars comprised in the group Parduca. The mean area values were comprised between 14.7 and 17.3, the mean Stummer index between 0.53 and 0.59, and the J index between 89.6 and 89.7.

Figure 14 contains an image of the model Parduca, an image with the silhouettes of 30 seeds of ‘Vijariego Blanco’, the average silhouette obtained by the pixel profile of the image with the silhouettes of ‘Vijariego Blanco’, and four representative seeds of the cultivar of ‘Vijariego Blanco’.

Seeds of cultivar ‘de Cuerno’ had mean values of 28.3 mm (área), 0.42 (Stummer index), and 89.3 (J index).

Table 7. Summary of results of the groups with Curvature Type 3. The values given correspond to minimum and maximum values among the means obtained for all the cultivars in each group (for details, see Supplementary Tables S8 and S9).

	Area	Stummer Index	J Index
Chardonnay	16.0–16.9	0.74–0.74	90.6–90.8
Parduca	14.7–17.3	0.53–0.59	89.6–89.7
de Cuerno	28.3	0.42	89.3

contains a summary of the results of the groups with Curvature Type 3.

3.4.4. Series of Groups with Intermediate or Undefined Peduncle End

This series includes a total of 53 cultivars in four groups: Sylvestris (23), Teta de Vaca (21), Airén (3), and Parraleta (3). The summary of the results of the groups with Curvature Type 4 are in

Table 8. Summary of results for the groups with Curvature Type 4. The values given correspond to minimum and maximum values among the means obtained for all the cultivars in each group (for details, see Supplementary Tables S10–S13).

	Area	Stummer Index	J Index
Sylvestris	11.8–20.1	0.58–0.75	89.6–91.9
Teta de Vaca	15.2–24.5	0.55–0.71	89.9–91.7
Airén	17.4–18.9	0.56–0.58	90.2–90.9
Parraleta	14.2–20.7	0.52–0.58	89.6–90.1

.

Supplementary Table S10 contains mean data for the area, Stummer index, and J index of the cultivars comprised in the group Sylvestris. The mean area values of the cultivars belonging to the group Sylvestris range between 11.8 and 20.1 mm². The mean Stummer index values range between 0.58 and 0.75. The mean J index values range between 89.6 and 91.9.

Figure 15 contains an image of the model Sylvestris and images with the silhouettes of 30 seeds, the average silhouette obtained by the pixel profile of the image with the silhouettes, and four representative seeds of cultivars ‘BA 2.3’, Se 2.7 y ‘Benedicto’.

Supplementary Table S11 contains mean data for the area, Stummer index, and J index of the cultivars comprised in group Teta de Vaca. The mean area values range between 15.2 and 24.5 mm². The mean Stummer index values range between 0.55 and 0.71. The mean J index values range between 89.9 and 91.7.

Figure 16 contains an image of the model Teta de Vaca and images with the silhouettes of 30 seeds, the average silhouette obtained by the pixel profile of the image with the silhouettes, and four representative seeds of the cultivars ‘Ohanes’, ‘Xaell,lo’ y ‘Gorgollasa’.

Supplementary Table S12 contains mean data for the area, Stummer index, and J index of the three cultivars comprised in group Airén. The mean area values range between 16.2 and 18.9 mm². The mean Stummer index values range between 0.56 and 0.58. The mean J index values range between 90.2 and 90.9.

Figure 17 contains an image of the model Airén, an image with the silhouettes of 30 seeds of ‘Mazuela’, the average silhouette obtained by the pixel profile of the image with the silhouettes of ‘Mazuela’, and four representative seeds of the cultivar of ‘Mazuela’.

Supplementary Table S13 contains mean data for the area, Stummer index, and J index of the four cultivars comprised in group Parraleta. The mean area values range between 14.2 and 20.7 mm². The mean Stummer index values range between 0.52 and 0.58. The mean J index values range between 89.6 and 90.1.

Figure 18 shows an image of the model Parraleta, images with the silhouettes of 30 seeds of ‘Sabaté’ and ‘Mondragón’, the average silhouette obtained by the pixel profile, and four representative seeds of cultivars ‘Sabaté’ and ‘Mondragón’.

3.5. Multivariate Analysis

The PCA with the distribution of 224 grapevine cultivars from Figure 19 shows different groups gathered to specific variables. Variables have a common distribution according to complex geometric models (Stummer Index, Circularity, and J index), direct measurements (Width, Length, Perimeter, and Area), and discrete values (Curvature). Two main variables gather most of the cultivars. Curvature is associated with the mean value of groups Airén, Cariñena Blanca, Doña Blanca, Parduca, Parraleta, and Rayada Melonera. On the other hand, the J index is connected to Albillo Real, Chardonnay, Hebén, Listán Prieto, and Silvestre. The distribution of Moscatel, Teta de Vaca, and Tortozón do not show a clear connection with any variable, and de Cuerno is an out layer for all variables.

A hierarchical clustering from the PCA using the mean value for the 15 groups provided a dendrogram (Figure 20). According to the branch color, the number of suggested clusters is three. Cluster 1 includes Tortozón, Listán Prieto, Moscatel, Albillo Real, Teta de Vaca, Hebén, Silvestre, and Chardonnay. Cluster 2 includes Airén, Rayada Melonera, Cariñena Blanca, Parduca, Doña Blanca, and Parraleta. de Cuerno, as an outlayer, has its own cluster.

4. Discussion

The classification of grape varieties according to the shape of their seeds as defined by geometric models has been the subject of two previous articles. Martín-Gómez et al. [35] applied eight models (M1 to M8) to the description and classification of seed shape in species of Cissus and Parthenocissus, two relatives of Vitis in the Vitaceae, as well as to V. amurensis, V. labrusca, V. vinifera subsp. rupestris, and to six cultivars of V. vinifera subsp. vinifera (‘Camarate’, ‘Cariñena’, ‘Cercial’, ‘Malvasía’, ‘Meserguera’, and ‘Morenillo’). Ten new models derived from M7 and M8 were obtained and used for the description and classification of 38 cultivars representative of the collection of IMIDRA [36]. In this article, the objectives have been expanded to the classification of a total of 224 cultivars of the IMIDRA collection, and for this, five new models derived from M7 have been described.

Stummer index (SI), the ratio width/length, was proposed as a measure useful for the distinction between grape cultivars on the basis that wild-type cultivars have higher values of SI. The quantification of the morphological similarity of the seeds with geometrical figures obtained from the representation of algebraic equations and used as models allows much more precision in the description of shape than the relationship between two measurements. Similarity with a given geometrical model shape description is independent of size and SI. Seeds with similar sizes or values of SI may resemble different geometric models, while seeds of different sizes may resemble the same model.

To choose the name of the representative variety of each group, the oldest cultivated variety in Spain with written reference was sought. The ‘de Cuerno’ variety (‘Datilera’) and ‘the Teta de Vaca’ variety are mentioned in the Book of Columela of the 1st century [42]. The varieties ‘Tortozón’, ‘Listán Prieto’ (‘Palomina Prieta’), ‘Moscatel’, ‘Albillo Real’ (‘Albillo’), ‘Hebén’, and ‘Doña Blanca’ (‘Cigüente’) appear in Alonso de Herrera’s work of 1513 [22]. He also mentions other varieties such as ‘Torrontés’, ‘Cayetana Blanca’ (‘Jaén’), ‘Alarije’, and ‘Tempranillo’ (‘Aragonés’), which are all included in the Albillo group, and the variety ‘Beba’ (‘Lairén’), which is included in the Moscatel group.

Analyzing the dendrogram in Figure 20, the ‘de Cuerno’ variety is a seed variety different from the rest, an ancient variety that possibly originated in the East or North Africa. In a second group, there are varieties cited in the 18th and 19th centuries by Valcarcel, Abela, and Clemente [43,44,45], together with the ‘Doña Blanca’ (‘Cigüente’) cited by Alonso de Herrera in the 16th century [22]. A third group contains the wild varieties, including those cited by Alonso de Herrera in the 16th century, ‘Tortozón’, ‘Listán Prieto’ (‘Palomina Negra’), ‘Moscatel’, ‘Albillo Real’ (‘Albillo’), and ‘Hebén’ [22]; the ‘Teta de Vaca’ variety of possible Egyptian origin and cited by Columela in the 1st century [42] and the ‘Chardonnay’ variety cultivated in Burgundy (France), where there is a locality with the name of the variety. The genetic origin of ‘Chardonnay’ comes from the crossing of ‘Gouais Blanc’ and ‘Pinoccio’, and it is of uncertain origin, either cultivated by the Romans or brought from Palestine by the Crusaders. Seed shape can be used to group varieties based on age with a high degree of confidence.

The discovery of clear morphological distinctions among cultivated vine varieties, which are often categorized into a limited number of models, can greatly facilitate the examination of archaeological remnants. Seeds, typically the most preserved organ of the vine in ancient remains, serve as tangible evidence of the origin and historical presence of these varieties. For instance, the wild vine is widely regarded as the progenitor of cultivated vines [16], and numerous studies have elucidated the morphological disparities in its seeds [19].

The vine has undergone spatial migration over time, as evidenced by its spread across ancient civilizations throughout the Mediterranean region. The domestication of grapevines from their wild counterparts occurred through various pathways, including the development of table grapes and wine grapes [16], as well as through introgressions with wild vines. These interactions may elucidate the morphological disparities observed between the seed forms of contemporary cultivated varieties and those of wild grapes.

The Wild model (Sylvestris in Table 2) includes 22 of the studied varieties, most of them red grape accessions, among which there are 14 wild accessions as expected, and minoritary varieties such as ‘Gonfaus’, ‘Zurieles’, ‘Benedicto’ (female flowering), ‘Lucomol’, and one of the most cultivated varieties in the world, ‘Merlot’. This variety derives from the crossing of ‘Cabernet Franc’ with an unknown variety. It was sampled for the first time in 1996 in an abandoned vineyard in Saint-Suliac, Brittany, which acted as the mother of ‘Merlot’. Our results support the idea that this could be a wild vine. However, not all wild accessions are clustered to this model; two of them fall into the groups of Doña Blanca from the South of France and Seville, one into that of Teta de Vaca from Navarra, one into that of Tortozón from Cádiz, and one into that of Moscatel from Córdoba. Some studies report diversity among wild populations [46], and although all the accessions analyzed in this work come from a similar geographical area, the Iberian Peninsula and the south of France, local environmental conditions may affect seed shape in Vitis.

The group with the largest number of varieties is the Moscatel model, with a total of 60 varieties, including all the varieties with muscatel flavor or special aromas (‘Muscat of Alexandria’, ‘Muscat of Angües’, ‘Moscatel de Grano Menudo’, ‘Moscatel Rosa’, ‘Malvasia Aromática’, ‘Gewürztraminer’, ‘Riesling’). It also includes many varieties that do not have this character, such as the group of Garnachas and even a ‘Silvestre’ variety from the area of Cordoba. The ‘Garnacha Blanca’ variety falls out of this group and shares the Doña Blanca model. This is an interesting fact because this variety arose because of a somatic mutation in the skin color, and yet there has been an additional change in the shape of the seed. Also, the aromatic variety ‘Italia’ shares a model with Doña Blanca and not with Moscatel, but this is not strange since it is not the offspring of ‘Moscatel de Alejandría’ [47].

The second-largest group comprises Doña Blanca, encompassing 45 varieties. Within this group, a considerable number originate from three distinct geographical regions, setting them apart from the rest due to their potential lack of Andalusian influence. These include Galician varieties (‘Albarín Blanco’, ‘Albariño’, ‘Brancellao’, ‘Castañal’, ‘Godello’, ‘Lado’, ‘Loureira’, ‘Mencía’, ‘Sousón’), some from France (‘Cabernet Franc’, ‘Folle Blanche’, ‘Sauvignon Blanc’, ‘Syrah’); and others from the Balearic Islands (‘Batista’, ‘Callet’, ‘Excursach’, ‘Fogoneu’, ‘Manses de Tibbus’, ‘Manto Negro’).

The Albillo Real model is followed by 27 varieties, including some of great importance worldwide (‘Cabernet Sauvignon’, ‘Tempranillo’, ‘Pinot Noir’).

Two other models with a high number of varieties are Hebén and Teta de Vaca, with 22 and 21 varieties, respectively. The Hebén model includes large cluster varieties, some of which are table varieties (‘Aledo’, ‘Aledo Real’, ‘Rosetti’). However, not all the varieties in this group are large-cluster varieties. ‘Aledo’, like ‘Hebén’, are female-flowered varieties.

The rest of the models cluster a low number of varieties and are, therefore, more selective for varietal identification, especially the de Cuerno model, which has only been found in this variety. The Chardonnay group includes only two varieties; the Listán Prieto group includes four varieties, as does the Parraleta group; the Airén, Cariñena Blanca, and Parduca groups include three varieties each; and the Tortozón group includes six varieties.

Consequently, we can say that according to the seed shape, the varieties cultivated in Spain generally group according to model similarity based on common characteristics. However, we can find some exceptions since the morphological characteristics of the seeds, like other morphological characteristics of the vine, derive from two progenitors and can resemble one of them or share a resemblance with both.

Let us scrutinize these classifications by scrutinizing the genealogical findings elucidated in prior research: ‘Malvar’, for instance, is the progeny of ‘Hebén’ and ‘Tortozón’ [45], hence, allocated within the Hebén group. Similarly, ‘Merseguera’ shares the same parentage but is classified under the Moscatel model. ‘Moscatel de Grano Menudo’, a forebear of ‘Moscatel de Alejandría’ alongside ‘Heptakilo’ [44], is also situated within the same model. In such instances, the seed morphology frequently mirrors inheritance from one of the parent cultivars.

‘Mandón’, conversely, stems from the union of ‘Hebén and Graciano’ [46], yet it falls into the Doña Blanca model, while its progenitors are categorized under the Hebén and Albillo Real models. ‘Moscatel de Angüés’, originating from ‘Moscatel de Alejandría’ and ‘Hebén’ [46], is likewise categorized within the Moscatel model.

‘Verdejo de Salamanca’ emanates from ‘Hebén’ and ‘Listán Prieto’, with each cultivar manifesting distinctive seed form patterns. Viura, an offspring of Hebén, adheres to the same shape model as Miguel de Arco [46]. This affirms that while seed morphology traits are heritable, they may not invariably align entirely with those of the parent cultivars.

5. Conclusions

Grapevine seed morphology is a useful tool for variety clustering. The study of models based on the seed shape properties of different varieties has proven to be a simple and effective tool for their classification. This technique can serve as a significant aid in identifying the materials cultivated in various regions in the past, as grapevine seeds are frequently unearthed in archaeological remains.

5.1. New Seed Models

Four new seed-shaped models appear in addition to the ten already shown in previous works. The new models complete the range of Stummer index (SI) with intermediate values between the 10 models described before, also completing the curvature series at the apex of the peduncle.

5.2. Distribution of All Cultivars Studied in the 14 Groups Defined by the Models

The total of 223 varieties studied fall into one of the 14 classification groups according to seed morphology.

5.3. Variation in Size and Shape in the Cultivars

The range of variation in seed dimensions such as length, width, S.I., area, and perimeter vary within the range reported in previous works. The smallest and most rounded seeds, such as those from wild vines (‘Sylvestris’ varieties), are supposed to be the oldest in history (less evolved) and the ones that have been least manipulated in the process of selection.

5.4. Morphological Groups by Series

The fourteen groups can be clustered in four series according to the shape at the end of the seed peduncle. The peduncle may end in an acute, plane, rounded, or intermediary-indeterminate shape.

5.5. Seed Shape Quantification in the Study of Varieties

In addition to known size and shape measurements, seed morphological measurements based on the comparison with geometric models are useful in the identification and classification of grape varieties.