Morphological Characterization of Opuntia Accessions from Tenerife (Canary Islands, Spain) Using UPOV Descriptors

: Twenty Opuntia accessions from Tenerife (Canary Islands, Spain) were classified according to 52 quantitative and qualitative descriptors, including the traits of the cladodes, flowers, fruits, and spines, as described by the International Union for the Protection of New Varieties of Plants (UPOV) guidelines. A database composed of 20 accessions and 52 traits was used to perform a cluster analysis based on the Euclidian distance and Ward’s method and a canonical discriminant analysis. In terms of the analyzed characteristics, cactus pears with orange flesh showed less variability than cactus pears with white or purple flesh. Good classifications according to fruit flesh color were obtained using discriminant analysis. As a result of the cluster analysis, Opuntia plant accessions with white, orange, or purple-fleshed fruits were classified into four homogeneous groups according to the cubic clustering criteria. This study proves that it was possible to make a preliminary classification of Opuntia varieties from the Canary Islands based on a few main morphological characteristics. To improve the classification, a molecular analysis of the different Opuntia plants is necessary.


Introduction
The genus Opuntia is a member of the succulent plant family Cactaceae [1,2].Cactaceae are dicotyledonous angiosperms that are most abundant in arid and semiarid regions, and they are the family with the largest number of genera, approximately 130, within the order Caryophyllales, which is predominantly distributed from northern Canada to Patagonia [3,4].Anderson [5] and Peña-Valdivia et al. [6] include approximately 181 species in the Opuntia genus, while some authors [7] estimate as many as 215 species.
The Opuntia genus is native to Mexico according to the great interspecific and intraspecific variability observed in the country [8].It is mainly present in North and South America and was imported to Europe by the first Spanish conquerors at the end of the 15th century, beginning in the 16th century, attracted by their morphology and the involvement of the cacti in the Aztecs economic, social and religious life [9].In 1568, Matthioli Sanese [10] stated that the environmental conditions of temperate Mediterranean areas controlled by the Spanish Empire (Sicily, southern Italy, and northern Africa) allowed its naturalization and spread to other countries, such as India, Thailand, and Australia [11,12].The cultivation was first performed in the Canary Islands (Spain) to produce a highly appreciated red dye obtained from the body of a cochineal insect (Dactylopius coccus Costa) housed in certain species of Opuntia [9].Similar to other cactaceaes, Opuntia species have an extraordinary capacity to store water [13] and grow in dry and even desert conditions [14].These plants have a crassulacean acid metabolism (CAM) that fixes carbon dioxide during the night when evapotranspiration is relatively low and stores it as organic acids to perform photosynthesis during the day when they close their stomata to avoid water loss.
Cacti were introduced for agriculture and ornamental horticulture in the Canary Islands a long time ago [15].Numerous plants from America were introduced and acclimatized in the Canary Islands during the 16th to 18th centuries since the islands were a strategic area on the American route during the first moments of the American conquest.Plants in the Opuntia genus are some of the most popular American plants established in the territory [16].In addition, as a result of the subtropical climate of the islands, numerous species are not only well adapted to be cultivated in the local climate but also indeed reproduce locally without human intervention, eventually naturalizing, spreading, or even becoming invasive.None of the cacti species are native to the Canary Islands; however, some of them currently occupy large areas of the island territory, such as Opuntia ficus-indica (L.) Mill.and O. dillenii (KerGawl.)Haw.O. dillenii is confined to coastal areas, while O. ficus-indica is omnipresent and it is located at very different altitudes [17].Opuntia plants are appreciated because of their considerable number of uses (fodder crop for animal feed, defensive hedges or important elements in erosion control and land rehabilitation, especially in arid and semiarid areas, and as a shelter and feed resource for wildlife, birds, and mammals, etc.) [18].In addition, fruits and fleshy stems, called nopales, have been used as food.Opuntia fruit, also known as cactus pear fruit, prickly pear, tuna (Mexico), higo (Colombia), higo chumbo (Spain), and figue de barbarie (France), are harvested from various species.The fruits can be consumed fresh, dehydrated, or in marmalades, while the nopal is consumed in Mexican regions, for example, as a constituent of salads [19].Moreover, there is growing interest in nonfood uses, especially in medical applications [20][21][22], in the cosmetic industry [23][24][25][26], in the production of carminic acid, which is used as a natural colorant [19], and in the area of agro-energy, which is used as a substrate in anaerobic digestion for the production of biogas and methane [27].
The cactus pear varieties and their wild relatives have been classified based on the morphological features of the fruits and cladodes [6,[28][29][30], chemical attributes [31,32], and frost tolerance [33].However, most of these studies have used only a few commercial varieties and a reduced number of features.On the other hand, some studies have used molecular markers such as isozymes [34,35], random amplified polymorphic DNA (RAPD) [36], chloroplast simple sequence repeats (cpSSRs) [37,38], and inter simple sequence repeats (ISSRs) [38].
The genus Opuntia includes a great genetic diversity, and its taxonomic classification has been considered very complex, which explains the existence of many reports of wrongly classifying Opuntia species [39].In fact, continuous morphological variation and limited morphological descriptors for cultivar discrimination are the most important obstacles to achieving stable classification [37,[39][40][41].In addition, farmers use different common names depending on the growing area, which makes it more complex to know the varieties used.
The present study aimed to document and classify 20 accessions of the Opuntia genus located in Tenerife (Canary Islands, Spain) using quantitative and qualitative descriptors, which include the traits of the cladodes, flowers, fruits, and spines as described by the International Union for the Protection of New Varieties of Plants [42] (UPOV, 2006) guidelines.

Plant Material
Twenty Opuntia accessions from different locations in Tenerife (Canary Islands, Spain) were evaluated for quantitative and qualitative traits.The plants were geolocated according to Figure 1, with their latitudinal and longitudinal coordinates.A total of 12 Opuntia accessions were cultivated by farmers, while the other 8 came from the cactus pear collection grown in the Center for the Conservation of Agricultural Biodiversity in Tenerife (CCBAT).Healthy plant material that was not affected by diseases or important diseases was selected for this study.Similarly, the selected material was free from any type of treatment that could affect the expression of the characteristics of the species [42].Ten individuals were scored, and from each individual, 10 cladodes, 20 fruits, 10 flowers, and 10 spines were evaluated.

Plant Material
Twenty Opuntia accessions from different locations in Tenerife (Canary Islands, Spain) were evaluated for quantitative and qualitative traits.The plants were geolocated according to Figure 1, with their latitudinal and longitudinal coordinates.A total of 12 Opuntia accessions were cultivated by farmers, while the other 8 came from the cactus pear collection grown in the Center for the Conservation of Agricultural Biodiversity in Tenerife (CCBAT).Healthy plant material that was not affected by diseases or important diseases was selected for this study.Similarly, the selected material was free from any type of treatment that could affect the expression of the characteristics of the species [42].Ten individuals were scored, and from each individual, 10 cladodes, 20 fruits, 10 flowers, and 10 spines were evaluated.The identification code, the common name provided by the farmer, and the location of the samples collected around Tenerife Island are summarized in Table 1.The farmers assigned the same common name to samples with different codes.The identification code, the common name provided by the farmer, and the location of the samples collected around Tenerife Island are summarized in Table 1.The farmers assigned the same common name to samples with different codes.

Morphological Descriptors
The measurements were performed according to the UPOV [42] (UPOV, 2006) guidelines.A total of 52 UPOV descriptors (Table 2), comprising 3 in the plant, 16 in the cladodes, 19 in the fruits, 5 in the flowers, and 9 in the spines, were measured and used to design a numbered-data matrix.The measurements were performed by the same person to avoid errors due to individual differences.

Chemical Parameters
The fruits of each individual were hand-peeled, minced, and homogenized for analysis in triplicate.Moisture was determined using the Official Method of Analysis of the Association of Official Analytical Chemists (AOAC) with the oven-drying method (P Selecta 207, Barcelona, Spain), and dry matter was calculated by difference [43] (AOAC 934.06).Total soluble solids (TSS) were determined using a hand refractometer (ATC-1, ATAGO, Tokyo, Japan) [43] (AOAC 932.12), pH was measured with an automatic titrator (Titralab AT1000, Berlin, Germany) [43] (AOAC 981.12), and total acidity (TA) (percentage of citric acid) was determined by titration with NaOH to an endpoint of pH 8.1 [43] (AOAC 942.15).

Statistical Analysis
Statistical analysis was performed using the statistical package SPSS 25.0 (SPSS Inc., Chicago, IL, USA).To detect and delete possible anomalous data, box plot data were explored.The Kolmogorov-Smirnov test was applied to assess whether the variables presented a normal distribution.Morphological descriptors were analyzed using linear discriminant analysis (LDA) and cluster analysis.LDA was carried out with the objective of differentiating and classifying the samples according to fruit flesh color.Hierarchical cluster analysis (HCA) was applied using Ward's method and the squared Euclidean distance within each of the preestablished groups according to flesh color (white, orange, or purple) to identify groups of accessions.

Morphological Analysis
Table 3 shows the morphological characterization of the seven Opuntia accessions with white-fleshed fruits.The width/height ratio (2.8 compared with an average of 1.8 for the remaining accessions) and length/width cladode ratio (2.9 vs. 2.1) were highest for the EGC009 accession.The color of the cladodes varied from light green to bluish-green.The length of the longest spine varied from large in the case of CBT02253 (29.0 mm) to small in the case of CBT02249 (9.4 mm), representing an intermediate size for the remaining accessions (mean value of 16.3 mm).The largest fruits were those of CBT02253, with an average length of 7.6 cm, a maximum diameter of 5.1 cm, and a flesh weight of 93.7 g, while CBT02826 presented the smallest fruits, with a maximum length and diameter of 2.9 cm. and 2.7 cm, respectively, and a flesh weight of only 5.3 g.The predominant shapes were medium elliptic and oblong, and the main skin color was light green, except in CBT02249 and CBT02826, which were orange and yellow, respectively.The flesh color was more homogeneous, varying between light green and medium green.Farmers used to call cactus pears with this flesh color "blancos", corresponding to white cactus pears.Among the orange-fleshed cactus pears (Table 4), the EGC006 accession had the highest width/height ratio (2.5 compared with an average value of 1.6).The length/width cladode ratio was very similar for all the accessions, with a mean value of 1.8.The cladodes of all the evaluated plants presented a light green color without pubescence on the surface or undulation of the margin.RTA014, EGC006, EGC008, EGC016, and EGC018 presented cladodes with a broad elliptic shape, while those of CBT02255 and RTA012 were narrow obovate and rhombic, respectively.The length of the longest spine ranged from 9.2 to 15.6 mm, with 12.6 being the mean value.The predominant color of the spines was white in all accessions, and the attitude of the central spine was semierect, with the exception of EGC018, which presented yellow spines and an erect attitude of the central spine.The perianth was yellow in all the orangefleshed accessions.The accession RTA012 presented the largest fruits, with a length of 7.4 cm and a maximum diameter of 6.0 cm.In general, all the fruits presented oblong shapes, except those of the accession EGC018, which were medium elliptic.The main skin color was orange, except in the accessions EGC006 and EGC018, which were medium green.Table 5 shows the significant variability in the descriptors studied among the six purple-fleshed accessions.The width/height ratio (1.9) was greater for the EGC005 accession than for the other accessions (mean value 1.2).The cladodes of EGC015 and RTA020 presented the highest length/width ratios (2.4 and 2.2, respectively).The shape of the cladodes was variable, and the color fluctuated from light to dark green.RTA020 and EGC026 presented surface pubescence, and the latter showed margin undulation as well.In addition, none of the accessions presented glochides, or the number of glochides was very low, except for accession ECG026, which presented a few yellow glochides.EGC026 presented the largest spines, with the longest spine being 56.0 mm, whereas accession CBT03000 had the smallest spine (5.9 mm).The shape of the central spine of EGC026 in the dorsal view was narrow triangular, while that of the remaining accessions was aciculate.CBT03000, RTA020, and EGC025 presented orange perianths, and ECG005, EGC015, and EGC026 presented yellow perianths.The fruits of EGC025 had the greatest length and maximum diameter (9.2 cm and 5.6 cm, respectively), as well as the greatest flesh weight (70.1 g).The smallest fruits were those of the RTA020 accession (3.5 cm in length, 3.3 cm in maximum diameter, and 12.1 g in flesh weight).The fruits were predominantly oblong and medium elliptical in shape.The skin color was purple except for that of the CBT03000 and RTA020 fruits, which were medium or dark red, respectively.Finally, all the accessions had purple flesh except CBT03000, which had pink flesh.

Chemical Composition Analysis
For white-fleshed fruits (Table 6), the dry matter (DM) content (10.80%),TSS content (7.93 • Brix), and pH (3.99) of accession CBT02826 were substantially lower, and the TA (0.91% citric acid) was much greater than those of the other accessions (15.88%, 14.23 • Brix, 6.87, and 0.03% citric acid, respectively).In the case of orange-fleshed cactus pears, accession RTA012 had the highest percentage of DM (18.37%), and EGC006 had the lowest percentage (12.37%).The TSS ranged from 12.69 • Brix (EGC018) to 16.83 • Brix (EGC008), while the mean pH and TA were 6.64 and 0.03% citric acid, respectively.Finally, for the purple-fleshed accessions, the TSS content (11.37 • Brix) and pH (3.29 compared with an average of 6.85) were lower in accession EGC026 than in the other accessions, and TA (1.00% compared with 0.03% citric acid) was much greater than that in the other accessions.However, their percentage of DM was similar to that of the purple-fleshed cactus pears (14.55%).

Linear Discriminant Analysis (LDA) and Hierarchical Cluster Analysis (HCA)
In the stepwise linear discriminant analysis (LDA) performed to differentiate the accessions by the color of their flesh, the percentages of correct group classification were 85.0% and 75.0% after cross-validation (Table 7), and three variables were selected to achieve this classification: skin color, cladode spine color, and flower perianth color.All purple-fleshed accessions were correctly classified; only one orange-fleshed accession was erroneously classified as white-fleshed, and two white-fleshed accessions were incorrectly classified as orange-fleshed.When performing the discriminant analysis by the direct method LDA (which includes all the studied variables), the percentages of correct classifications were 100% and 55.0% after performing the cross-validation.Two estimated canonical functions explained all the variability; the first accounted for 99.9% of the total variance, and the second accounted for 0.1% (Figure 2).After performing the LDA, a ward-derived dendrogram was constructed considering the flesh color of the cactus pears and based on the 52 UPOV descriptors evaluated.Figure 3 separates the seven white-fleshed accessions into three main groups, and the most frequent characteristics of the groups are shown in Figure 4.The first, including the code CBT02826 (Group A), is characterized by many large spines and a very small fruit compared with the rest of the accessions.The second group also included a single accession, CBT02253 (Group B), with abundant and large spines but a larger fruit.The last group (Group C) was divided into two groups.One included one accession (EGC001, Subgroup C.1), and the other comprised the codes EGC019, CBT02249, EGC011, and EGC009 (Subgroup C.2). EGC001 was characterized by elongated and narrow cladodes and a flower with an orange perianth.However, the other subgroups presented more heterogeneous characteristics, as shown in Figure 4.In general, the cladodes were round, the flower perianths were yellow or orange, and the size of the fruits was intermediate.After performing the LDA, a ward-derived dendrogram was constructed considering the flesh color of the cactus pears and based on the 52 UPOV descriptors evaluated.Figure 3 separates the seven white-fleshed accessions into three main groups, and the most frequent characteristics of the groups are shown in Figure 4.The first, including the code CBT02826 (Group A), is characterized by many large spines and a very small fruit compared with the rest of the accessions.The second group also included a single accession, CBT02253 (Group B), with abundant and large spines but a larger fruit.The last group (Group C) was divided into two groups.One included one accession (EGC001, Subgroup C.1), and the other comprised the codes EGC019, CBT02249, EGC011, and EGC009 (Subgroup C.2). EGC001 was characterized by elongated and narrow cladodes and a flower with an orange perianth.However, the other subgroups presented more heterogeneous characteristics, as shown in Figure 4.In general, the cladodes were round, the flower perianths were yellow or orange, and the size of the fruits was intermediate.The dendrogram of orange-fleshed cactus pears (Figure 5) classified the seven accessions studied into three groups.Group A included only RTA012, which had differentiated cladodes from the other accessions (with a rhombic shape) and fruits with yellower flesh.Group B was divided into two groups: subgroup B.1, which included only EGC006, with cladodes with wax and larger fruits, and subgroup B.2, which included EGC018 and EGC016, with smaller fruits.Group C included the three remaining accessions (CBT02255, EGC008, and RTA014).The flowers of these three accessions had a white style, and their fruits were intermediate in size compared with those of the other groups.As shown in Figure 6, no notable morphological differences were detected between the established groups of orange-fleshed cactus pear accessions, which is in accordance with the results of the morphological characterization presented in Table 4. Therefore, and especially in this case, the difficulty of carrying out a characterization based on morphological characters is clear, and it is necessary to carry out a molecular analysis.The dendrogram of orange-fleshed cactus pears (Figure 5) classified the seven accessions studied into three groups.Group A included only RTA012, which had differentiated cladodes from the other accessions (with a rhombic shape) and fruits with yellower flesh.Group B was divided into two groups: subgroup B.1, which included only EGC006, with cladodes with wax and larger fruits, and subgroup B.2, which included EGC018 and EGC016, with smaller fruits.Group C included the three remaining accessions (CBT02255, EGC008, and RTA014).The flowers of these three accessions had a white style, and their fruits were intermediate in size compared with those of the other groups.As shown in Figure 6, no notable morphological differences were detected between the established groups of orange-fleshed cactus pear accessions, which is in accordance with the results of the morphological characterization presented in Table 4. Therefore, and especially in this case, the difficulty of carrying out a characterization based on morphological characters is clear, and it is necessary to carry out a molecular analysis.The dendrogram (Figure 7) of the purple-fleshed cactus pears classified the six studied accessions into two main groups.The first group (Group A) included plants that produced small pink fruits (average length of 45.7 mm and diameter of 33.9 mm).This group was divided into two subgroups: EGC026 (Subgroup A.1), CBT03000, and RTA020 (Subgroup A.2).The accession EGC026 presented yellow flowers, cladodes with many large spines and fruits with purple skin and flesh.The second group (Group B) included accessions with larger fruits (average length of 76.8 mm and diameter of 53.3 mm) and was split into two subgroups: EGC025 (Subgroup B.1) and EGC015 and EGC005 (Subgroup B.2).The accession EGC025 was differentiated from Subgroup B.2 because the cladodes were more rounded, thicker and green darker, and the flowers with orange perianths and fruits were larger.There were notable differences among the different groups of purple-fleshed cactus pear accessions (Figure 8) at the level of all structures: cladodes, flowers, style and stigma, and fruits.The dendrogram (Figure 7) of the purple-fleshed cactus pears classified the six s ied accessions into two main groups.The first group (Group A) included plants that duced small pink fruits (average length of 45.7 mm and diameter of 33.9 mm).This g was divided into two subgroups: EGC026 (Subgroup A.1), CBT03000, and RTA020 ( group A.2).The accession EGC026 presented yellow flowers, cladodes with many l spines and fruits with purple skin and flesh.The second group (Group B) included a sions with larger fruits (average length of 76.8 mm and diameter of 53.3 mm) and was into two subgroups: EGC025 (Subgroup B.1) and EGC015 and EGC005 (Subgroup The accession EGC025 was differentiated from Subgroup B.2 because the cladodes more rounded, thicker and green darker, and the flowers with orange perianths and f were larger.There were notable differences among the different groups of purple-fle cactus pear accessions (Figure 8) at the level of all structures: cladodes, flowers, style stigma, and fruits.

Discussion
Considering the morphological descriptors, white-fleshed fruits showed high variability in the traits analyzed for cladodes and fruits among the plants.However, the traits of orange-fleshed cactus pears were less variable than those of white-fleshed cactus pears.The six purple-fleshed accessions differed mainly in terms of descriptors related to cladodes, spines, and fruits.These results were similar to those of the chemical composition of the fruits since no major differences were found among the accessions of the orange-fleshed fruits.However, in white pulp fruits and purple-fleshed fruits, the accessions CBT02826 and EGC026, respectively, presented low DM, TSS, and pH values and high TA values in comparison with the other accessions.
Hadjkouider et al. [30] reported that only a few descriptors have an effective discriminating capacity.In their study, two quantitative descriptors (length of flower and length of the longest spine per cladode) and six qualitative descriptors (time of beginning of flowering, time of harvest maturity, main color of the spine, main color of the fruit surface, color of the fruit flesh, and firmness of the fruit flesh) were identified as differential parameters in Opuntia ficus-indica, Opuntia amyclaea, Opuntia streptacantha, Opuntia robusta, and Opuntia engelmannii.Chalak et al. [14] reported that only five variables were sufficient to discriminate among forty-three accessions (O.ficus-indica): fruit weight, peel weight, pulp weight, juice weight, and peel thickness.
In our study, only three variables (skin color, cladode spine color, and flower perianth color) of all those studied were selected to classify (85% stepwise LDA) cactus pears according to flesh color.According to the cluster analysis, once the flesh color was fixed, the descriptors with the highest discriminative capacity for white, orange, and purple-fleshed cactus pears were fruit size and cladode shape.In the case of white and purple cactus pears, the size and density of the spines, as well as the color of the perianth, were also discriminative.For orange-fleshed cactus pears, the waxiness of the cladodes and the color of the flower style were additional discriminative descriptors.For purple-fleshed fruits, the thickness and color of the cladodes, as well as the color of the skin and pulp of the fruits, were other descriptors that allowed differentiation of the accessions.
These results may contribute to the best management of genetic diversity in Opuntia species from the Canary Islands through the establishment of a main collection that includes the most representative species with significant variation in the descriptors.This collection would benefit the conservation and breeding programs of Opuntia species without resorting to molecular techniques, which are much more complex and expensive.

Figure 1 .
Figure 1.Geolocation of the 20 selected Opuntia accessions on Tenerife Island.

Figure 1 .
Figure 1.Geolocation of the 20 selected Opuntia accessions on Tenerife Island.

Figure 2 .
Figure 2. Representation of the first two canonical discriminant functions differentiating by the flesh color of fruits.

Figure 2 .
Figure 2. Representation of the first two canonical discriminant functions differentiating by the flesh color of fruits.

Figure 3 .
Figure 3. Dendrogram showing relationships between seven white-fleshed Opuntia accessions using 52 UPOV descriptors based on the squared Euclidean distance and Ward s method.

Figure 3 .
Figure 3. Dendrogram showing relationships between seven white-fleshed Opuntia accessions using 52 UPOV descriptors based on the squared Euclidean distance and Ward´s method.

Horticulturae 2024 ,Figure 3 .
Figure 3. Dendrogram showing relationships between seven white-fleshed Opuntia accessions usin 52 UPOV descriptors based on the squared Euclidean distance and Ward s method.

Figure 5 .
Figure 5. Dendrogram showing relationships between seven Opuntia orange-fleshed accessions using 52 UPOV descriptors based on the squared Euclidean distance and Ward s method.

Figure 5 .
Figure 5. Dendrogram showing relationships between seven Opuntia orange-fleshed accessions using 52 UPOV descriptors based on the squared Euclidean distance and Ward´s method.

Figure 7 .
Figure 7. Dendrogram showing relationships between seven purple-fleshed Opuntia accessions using 52 UPOV descriptors based on the squared Euclidean distance and Ward´s method.

Figure 7 .
Figure 7. Dendrogram showing relationships between seven purple-fleshed Opuntia accessio ing 52 UPOV descriptors based on the squared Euclidean distance and Ward s method.

Table 1 . Code, common name, and localization of the 20 Opuntia accessions used for the characterization. Code Common Name Localization
SN: Without a typical common name.

Table 6 .
Chemical characterization of white, orange, and purple cactus pears.

Table 7 .
Classification of the 20 accessions based on fruit flesh color by the stepwise method (LDA).

Table 7 .
Classification of the 20 accessions based on fruit flesh color by the stepwise method (LDA).