Materials and Methods

Plant material. Five Opuntia species of the subgenus Platyopuntia (Engelm.) Kreuz, Opuntia ficus-indica, Opuntia amycleae Ten., Opuntia engelmannii Salm-Dyck ex Engelm., Opuntia streptacantha Lem., and Opuntia robusta H.L.Wendl., were prospected from the Algerian steppes in Doukkara (Tébessa), Belaiba (M’sila), Mesrane (Djelfa) and Choucha (Laghouat) districts (Figure 1). In our study, we pre-considered O. amycleae as separate sepecies from O. ficus-indica just to check the morphological similarity. However, several authors consider O. amyclaea as synonyms of O. ficus-indica, and they divided this latter species into two botanical forms: a) O. ficus-indica f. amyclaea (Ten.) Schelle, with presence of spines and b) O. ficus-indica f. inermis (Web.) (^{Kiesling 1999}), which is spineless. In the present work we adopted a random sampling method; the sampled species were then kept at the repository of the High Commissariat for Steppes Development (HCSD) and National Institute of Forestry Research (NIFR) in Algeria (Table 1).

Figure 1 Algerian Opuntia species investigated in this study: 1: Opuntia ficus-indica; 2: Opuntia amycleae; 3: Opuntia robusta; 4: Opuntia streptacantha; 5: Opuntia engelmannii. 

Species analysis per species was applied to assess morphological variation, 10 individuals were scored, and from each individual 3 cladodes, 3 flowers and 3 fruits were investigated for their differences according to the 49 UPOV morphological descriptors (Table 2).

Morphological descriptors and data collection. Forty-nine major quantitative and qualitative traits, selected from the UPOV list (^{UPOV, 2006}), comprising cladode, flower and fruit descriptors were measured and used to design a numbered-data matrix (Table 2). Ten individuals per species were evaluated for their morphological diversity across the 49 descriptors. Measurements were performed by the same two persons to avoid errors due to individual variations.

Statistical analysis based on the morphological diversity of cladode, flower and Opuntia fruits. Morphological and phenological data were analyzed by multivariate analyses and clustering using XLSTAT software (Addinsoft, http://www.xlstat.com). PCA was performed to identify species groups and to determine the axes and the characters significantly contributing to the variation. In this procedure, the similarity matrix was used to generate eigenvalues and scores for the species. The first two principal components, which accounted for the highest variation, were then used to plot two-dimensional scatter plots. Hierarchical cluster analyses (HCA) was carried out using Ward’s minimum variance method as a clustering algorithm (^{Williams 1976}) and squared Euclidean distances as a measure of dissimilarity (^{Ward 1963}).

Results

Principal Component Analysis (PCA). The Principal Component Analysis (PCA) showed a high level of morphological variation between the five studied Opuntia species (Table 3). The first 2 axes of the PCA accounted for 67.217 % of the total variation with the first component (PC1) explaining 38.239 % of the total variation, while PC2 accounted for 28.978.

Out of 49 morphological descriptors, 8 were identified as the most discriminating and useful for the classification of the Opuntia species. The contribution of all parameters to the definition of the first two PCA axes is given in Table 3. The traits that best contributed to the first component variability were the following: time of beginning of flowering (0.229), time of harvest maturity (0.229), flower length (0.203), main spine color (0.191), main color of fruit surface (0.186), color of fruit flesh (0.186), firmness of fruit flesh (0.186), length of the longest spine per cladode (0.182). On the other hand, in the second PCA component, we obtained highly and positive correlation for the descriptions cladode width (0.217), cladode thickness (0.239), cladode color (0.238), color of areoles per cladode (0.210), fruit shape (0.197), peel weight per fruit (0.264), flesh weight per fruit (0.212), and number of fully developed seeds per fruit (0.181) (Table 3, Figure 2).

Figure 2 Principal Component Analyses showing: A) the contribution of the UPOV descriptors (from 1 to 49, see Table 2) to the obtained morphological variation; B) the distinction of all studied Opuntia species. 

The projection of the species on the plot delimited by the first two PCA axes, showed the classification of the species into four distinct groups (Figure 2). The first group (a) comprises Opuntia engelmannii a second group (b) includes two species, Opuntia ficus-indica and Opuntia amycleae situated at the negative side of the plot, Opuntia robusta constituted the third group (c) and Opuntia streptacantha constituted the fourth group (d).

Cluster Analysis. To understand how the 5 Opuntia species are structured a ward-derived dendrogram was constructed based on the data yielded by the 49 morphological descriptors (Figure 3). This allowed the identification of 3 groups, unlike the PCA and phenological that identified 4 groups (Figure 3). The first cluster (C1) was constituted by three species: Opuntia robusta (OR), Opuntia ficus-indica (OFI) and Opuntia amycleae (OA). Within this group, Opuntia robusta was very different and characterized by a high number of fully developed seeds per fruit, broad thick bluish green cladodes, with black areoles, oblate fruit shape, medium peel weight and heavy fruit flesh weight (Figure 3). Opuntia ficus-indica and Opuntia amycleae were the closest as both were characterized by brown glochides, medium flower length, pink flower style, green stigma lobes, a medium ratio of length/maximum diameter of fruit, dense areoles, medium diameter of the receptacle scar.

Figure 3 Dendrogram showing relationships between five Opuntia species in Algeria using 49 morphological descriptors (G1 and G2 stand for the identified groups). 

Opuntia streptacantha formed the second cluster (C2) having an obovate cladode shape, two types of color spines, a horizontal attitude of the central spine, orange colored perianth, high number of stigma lobes and glochides.

The third group (C3) was constituted by Opuntia engelmannii (OE) that is distinguishable particularly, by late time of beginning of flowering and harvest maturity, long flower length, yellow spine color, medium red fruit surface, red fruit firm flesh, high number of long spines per areole per cladode, and large amount of many glochides (Figure 3).

Discussion

The present study revealed the morphological and phenological diversity in the five dominant species of Opuntia grown in Algerian steppes using 49 morphological descriptors.

The set of the UPOV descriptors used here allowed, for the first time, the characterization of the morphological diversity in 5 Algerian Opuntia species. The number of the descriptors was high if compared to previous studies in Cactaceae with 24 descriptors: seven describing the cladode (length, width, thickness, ratio length/width, number of areoles in the central row, number of spines per areole, length of the longest spine), three for the flower (number of flowers per cladode, length of the flower, number of stigma lobes) and 14 related to the fruit (length, maximum diameter, ratio length/maximum diameter, density of areoles, peduncle length, depression of receptacle scar, receptacle diameter, peel thickness, peel weight, pulp weight, ratio weight of peel/fruit weight, number of normal seeds, number of abortive seeds, total soluble solids (°Brix) in ^{Gallegos-Vázquez et al. (2011)}, 27 descriptors from cladodes, flowers and fruits were measured in ^{Gallegos-Vázquez et al. 2012} and 23 descriptors in ^{Mejía et al. (2013)}.

According to the analysis performed, only few descriptors have shown an effective discriminating capacity. Two quantitative descriptors (length of flower and length of the longest spine per cladode,) and six qualitative descriptors (following: time of beginning of flowering, time of harvest maturity, spine main color, main color of fruit surface, color of fruit flesh, firmness of fruit flesh) have been identified as differential parameters in Opuntia ficus-indica, Opuntia amyclaea, Opuntia streptacantha, Opuntia robusta and Opuntia engelmannii (Table 3). These descriptors ranked within the primary list established by the UPOV for the characterization of Opuntia species (^{UPOV, 2006}).

In comparison with other results, five variables were found to be significant to discriminate forty three accessions (Opuntia ficus-indica) fruit weight, peel weight, pulp weight, juice weight and peel thickness were considered as the most discriminative traits (^{Chalak et al. 2014}), Three variables were selected as significant to discriminate twenty nine cactus pear varieties and four xoconostle varieties: peel thickness, pulp weight and to a lesser extent the diameter of the receptacle were the most discriminating of the UPOV guidelines traits (^{Gallegos-Vázquez et al. 2011}). Moreover, spine number, margin hardness and areole distance were selected the three major variables to discriminate between three populations of Hylocereus Britton & Rose spp. (^{Mejía et al. 2013}). In addition, ten morphological descriptors including; recepticular scar diameter, seed number, pulp weight, cladodes shape, spines, fruits number, fruits size, recepticular scar position, peel and pulp color were registered to be significant to discriminate Opuntia ficus-indica and Opuntia amyclaea (^{Erre & Chessa 2013}).

It has been reported that phenotypic differences between populations can be linked to environmental and genetic variables (^{Muñoz-Urias et al. 2008}). Additionally, ^{Reyes-Agüero (2005)}, reported the elements of the syndrome of domestication in Opuntia and many traits are of primary interest for human utilization and have had a clear effect on domestication (^{Gallegos-Vázquez et al. 2012}). The use of morphological descriptors comprising cladode, flower and fruit traits yielded a high number of morphotypes and allowed the discrimination of all of the studied species. In previous studies, the discrimination level was found to be as high as those reported by other studies using molecular markers for Opuntia spp (^{Labra et al. 2003}, ^{García-Zambrano et al. 2006}, ^{Zoghlami et al. 2007}, ^{García-Zambrano et al. 2009}, ^{Caruso et al. 2010}, ^{Majure et al. 2012}, ^{El-Banna et al. 2013}, ^{Valadez-Moctezuma et al. 2014}, ^{Ganopoulos et al. 2015}, ^{Samah et al. 2016}).

Multivariate analyses based on morphological characters are continuously providing valuable information allowing the breeder to improve the species selected from specific geographical regions (^{Souza & Sorrells 1991}). Previously, the multivariate techniques have been applied in Opuntia aiming at the morphological and the agronomic descriptions (^{Reyes-Agüero et al. 2005}, ^{Peña-Valdivia et al. 2008}) and the industrial (^{Hammouch et al. 2013}).

In our case, multivariate analyses have shown that the highest amount of variation was attained using fruits descriptors (Table 3). These were previously shown as the most important characters for the description of the species (^{Valdez-Cepeda et al. 2003}, ^{Reyes-Agüero et al. 2005}, ^{Gallegos-Vázquez et al. 2012}).

Morphological descriptors were identified as powerful tools in providing meaningful data on the qualitative and the quantitative aspects in some Opuntia species. Morphological characterization is needed to provide the users with valuable information on individual accessions, relationship among traits, and the structure of collections.

In this study, we managed to provide insights on the level of polymorphism and the structuring of the diversity in five Opuntia species from the Algerian steppes using 49 UPOV morphological and phenological descriptors.

These results may contribute to the best management of the genetic diversity in Opuntia species from Algeria throughout the establishment of a core-collection sheltering representatives from interesting species that best exhibit variation, and this would be a benefit for conservation and for future breeding programs.