Study area

The research was conducted in the northern and central regions of the state of Guerrero; specifically, in four municipalities: 1) Huitzuco de los Figueroa; 2) Atenango del Río; 3) Chilapa de Álvarez; and 4) Ahuacuotzingo; in which eight locations were studied: 1) Atetetla; 2) Paso Morelos; 3) Coacan; 4) Los Amates; 5) Ayahualco; 6) Santa Cruz; and 7) Motuapa and old Trapiche (Figure 1).

Figure 1 Geographical location of the study area, municipalities of Huitzuco de los Figueroa, Atenango del Río, Ahuacuotzingo and Chilapa de Álvarez, Guerrero, Mexico.  

The selection of localities was made based on the existence of maguey populations, in their wild form (Los Amates locality) or cultivated (all others), are localities that produce mezcal. They were georeferenced with respect to their latitude, longitude and altitude of each sampling site, it was carried out with a global positioning system (GPS), Garmin brand, model eTrex 30. The altitude ranged between 800 and 1 500 m (Table 1).

Phenotypic descriptions of the maguey plant were carried out (n= 57). In the sampling sites, the technical guide was used for the varietal description of Agave spp. (^{SNICS, 2014}). 32 variables were evaluated for each specimen, the description of the variables measured in: plant, stem, leaf and suckers of each maguey are:

From the 57 determined specimens, 1 824 data were obtained, with which a homogenization of the 32 variables evaluated was carried out based on the technical guide (^{SNICS, 2014}). The data were subjected to Pearson correlation analysis with significance level of p= 0.05, as a measure to identify the relationship or association between randomly selected variables. According to ^{Castañon-Najera et al. (2008)}, the variables that contribute little or nothing to the explanation of the two-dimensional plane were eliminated.

Therefore, the variables habit of plant growth was eliminated; number of leaves per filotaxy, visibility of the stem, width, curvature, secondary color, shade of the second color and distribution of the second color of the leaf, lateral spines, profile of lateral spines, prolificacy of shoots and cycle at the beginning of flowering. Subsequently, the principal components analysis (PCA) was carried out. The ACP aims at the transformation within a system of relevant coordinates and the reduction of the dimension. With this analysis the correlated variables are taken, which describe observations and a linear combination is found to generate new uncorrelated variables, called main components (CP or PC), so the ACP was generated with the 20 variables that provided the most explanation.

That is, 12 variables with very low significance and high correlation values were discarded. This reclassification of variables was made based on the correlation, with the purpose that the variables involved in the analysis had the same importance (^{Crossa et al., 1995}). The significance of the eigenvalues was determined by the Keizer rule (1960). An agglomerative hierarchical clustering analysis (AJA) was also performed, using the UPGMA linkage method, the measure of dissimilarity was the Euclidean distance. Statistical analyzes were performed with the XLSTAT Software (^{Fahmy, 1993}).

Principal component analysis (PCA)

In Table 2, it is shown that the first six CPs explain 79% of the total variation in the data. Therefore, they generate basic knowledge about the relationships between variables and main components. The CP1, explains 30% of the total variation with eigenvalue of 5.976, while the CP2 22% of the variation with eigenvalue of 4.385. CP3, CP4, CP5 and CP6 explained 9.647, 6.774, 5.874 and 4.874% of the morphological variation with eigenvalues of 1.929, 1.355, 1.175, and 0.975, respectively.

In the two-dimensional plane (Figure 2), group I is made up of specimens of A. angustifolia ‘sacatoro’ and group II, by specimens of A. angustifolia Haw ‘espadín’ with positive correlation between them.

Figure 2 Two-dimensional plane of CP1 and CP2. Group I (maguey sacatoro). Atetetla (HAT); Paso Morelos (HPM); Coacan (ARC); The Amates (CLA); Motuapa (AMO) and Trapiche Viejo (ATV). Group II (maguey espadín). Santa Cruz (SCS). Group III (maguey papalote). Ayahualco (CAY).  

That is, groups I and II have a strong correlation, so it tends to be located and grouped towards the left end of CP2 in the two-dimensional plane and group III formed by specimens of A. cupreata Trell & Berger ‘papalote’, sample a positive correlation between specimens, but negative with respect to groups I and II.

The CP1, was determined in a positive and significant way by the shape, edge and color of the leaf, striae on the lateral spines, distance between lateral spines and length of the terminal spine and negatively influenced the number of lateral spines, length of the leaf and the height of the plant. The CP2, is determined positively and significantly by the glaze of the leaf, uniformity in the size of the thorns, color of the lateral spines, number of leaves and shape of lateral spines, in the same component influenced negatively the texture of the leaf and the number of lateral spines.

The CP3, is formed by diameter of the rosette, height of the plant and length of the leaf, while the CP4, CP5 and CP6 are formed by a single variable, relationship between length and width of the leaf, shape of lateral thorns and intensity of the leaf color, respectively. It is worth mentioning that only the graphical results of CP 1 and 2 are presented, because it is visually complicated to understand the analysis with the 6 CPs.

^{Mora-López et al. (2011)}, took into account quantitative variables such as height and diameter of the rosette, length of the leaf, number, length, and distance between the lateral spines, which allowed them to separate populations of the genus Agave. ^{Rivera-Lugo et al. (2018)} in its research on the morphological and genetic variation of the A. angustifolia complex, showed useful traits for the taxonomic identification of this species and shows, with a phenotypic dispersion, that the taxa are clearly grouped according to the level of cultivation and management that is it gives to the plant. However, qualitative variables, although they are little influenced by environmental conditions, are also determinant (^{Franco and Hidalgo, 2003}).

Although A. angustifolia ‘sacatoro’ and A. angustifolia ‘espadín’ show a positive correlation with each other, it is clear that maguey espadín is distinguished by its greater domestication. This distinction of this species is due to the fact that it shows larger leaves and higher plant height, in contrast to the results, taxonomic implications of the morphological and genetic variation of the populations of the A. angustifolia de ^{Rivera-Lugo complex (2018)}, in this work was not considered as a reproductive stage, but in the field trips it was possible to corroborate that the inflorescence and terminal spine compared to the maguey sacatoro is larger. ^{Rodríguez-Garay et al. (2009)}, determined with an ACP study the relationship between A. tequilana var. Azul and A. angustifolia var. Lineño, obtaining a total variability 79% similar to that obtained in this study using the same technique, but with A. angustifolia ‘sacatoro’ and ‘espadín’.

^{Figueredo et al. (2014)} through multivariable methods, based on morphological characters, differentiated populations of wild and cultivated maguey reflecting artificial selection; showing that A. hookeri could be the end of a domestication gradient in a species complex, due to the similarity it presents with cultivated A. inaequidens. On the contrary, wild magueys such as A. macroculmis are small plants with large spines. Taking into account that the mechanical protection of the lateral spines is an important characteristic in the plant towards its predators, due to the incipient management conditions in which it is found (^{Castañon-Najera et al., 2008}).

In the same way, the maguey sacatoro, collected in the locality Los Amates in its wild form, has a small rosette (80 cm high) and small and numerous lateral spines (more than 150 lateral thorns) in comparison with the specimens evaluated in the Atetetla, Paso Morelos and Motuapa localities with an incipient management, which have a plant height of more than two meters, large lateral spines and in smaller quantity (from 60 to 80 lateral thorns in the leaf).

That is why, when a population of specimens is characterized in a morphological way and we work with quantitative and qualitative variables, it is important to reclassify variables based on correlation or multivariate normality tests, applying the central limit theorem, it is understood that samples comprise a ‘normal’ population (^{Castañon-Nájera et al., 2008}). The distribution and grouping of the specimens were complex; therefore, it was decided to consider a multidimensional analysis to form classes or groups defined by similarities as well as by morphological dissimilarities (^{Figueredo et al., 2014}; ^{Avendaño-Arrazate et al., 2015}).

Analysis of agglomerative hierarchical clustering (AJA)

The agglomerative hierarchical clustering analysis (AJA) represents the way in which the algorithm groups individuals from the Euclidean distance and then the subgroups, from the cofenetic distance. Finally, the algorithm regroups the observations progressively. The dotted line presents the truncation and shows the grouping defined by the Euclidean and cofetic distances in three groups, represented by maguey sacatoro (group I), maguey espadín (group II) and maguey papalote (group III).

Group III is shaped by a maguey papalote in a rearrangement of the subgroups with a cofeetic distance of 317. Group II is formed by maguey espadín, at a distance of 219 and the group I, consisting of maguey sacatoro. Although group I is well differentiated from group II and III at a dissimilarity distance of approximately 75%, it is divided into at least two subgroups; subgroup A, at a cofeetic distance of 35, consisting of 16 specimens: ARC1, HAT3, HAT5, AMO2, CLA4, CLA5, ARC6, AMO1, ARC2, ARC7, HPM1, ARC5, HPM5, ARC10, ARC4 and CLA1 and the subgroup B, at a cofeetic distance of 49 made up of 21 specimens: HPM3, HPM2, CLA7, HAT2, ATV3, ARC8, HPM4, ATV4, ATV5, CLA6, ARC9, CLA2, AMO4, AMO5, HAT1, CLA3, ATV1, ARC3, AMO3, HAT4 and ATV2, which confirms the grouping of the ACP (Figure 3).

Figure 3 Dendrogram of maguey specimens in three groups: A. angustifolia Haw (maguey espadín), A. angustifolia (maguey sacatoro) and A. cupreata (maguey papalote).  

The specimens of A. angustifolia sacatoro, although they are located in a separate subgroup, are also part of group II that makes up A. angustifolia ‘espadín’. Said sub-grouping is due to the type of management of the plant. The agaves evaluated in the field and some of their characteristics with which the ACP analysis and the AJA analysis were performed are observed in the (Figure 4 and 5).

Figure 4 Taxon A. angustifolia ‘psacatoro’ and sampled localities: A) Trapiche Viejo; B) Atetetla; C) Motuapa; and D) The Amates.  

Figure 5 Taxa and localities sampled. A) A. angustifolia ‘sacatoro’, locality Paso Morelos. B) A. angustifolia ‘papalote’, locality Ayahualco; C) A. angustifolia ‘sacatoro’, locality Coacan; and D) A. angustifolia ‘espadín’, locality Santa Cruz.  

The analysis of the main components and hierarchical grouping were useful tools for grouping the specimens according to the similarities or dissimilarities of the morphological variables of the agaves studied. In addition, in other investigations, using quantitative and qualitative characteristics through ACP analysis, the morphological and genetic relationships of this species have been identified (^{Colunga-García et al., 1996}; ^{Castro-Castro, 2010}). However, according to ^{Rodríguez-Garay et al. (2009)}, a genetic analysis, would yield more information about the morphological variation among the populations evaluated.

The results of the evaluated variables served to determine the morphological characteristics of the taxon A. angustifolia ‘sacatoro’. According to ^{Narez-Jiménez et al. (2014)}, the demand for raw material to be able to maintain the production of mezcal in the cultivated plants indicates that the magueyes constitute a valuable genetic resource that deserves to be studied to improve its use and conservation.