Introduction

The cultivation of chili is one of the most important horticultural crops in the world (^{de la Cruz-Lázaro et al., 2017}). It is distinguished by the pungency of its fruits, mainly caused by three capsaicinoids: nordihydro-, dihydro- and capsaicin (^{Sánchez-Sánchez et al., 2010}). It presents a wide variability of shapes, colors, aromas, flavors and sizes that constitute a valuable contribution of Mexico to the culture, food and gastronomy of the world since pre-hispanic times (^{Alonso et al., 2012}; ^{Hernández-Verdugo, 2014}; ^{Pérez et al., 2015}), with great relevance in Mexican cuisine, being the main spice (^{Zegbe et al., 2012}).

The chemical-food value is important because of the content of valuable antioxidant compounds in the prevention of chronic and degenerative diseases (^{López-López et al., 2015}). Likewise, its economic value is increased in the process industry and extraction of dyes and oleoresins (^{Arreguin et al., 2015}). In Mexico, the production of chili is 1 276 123 t ha^-1, the main producing states are: Sinaloa, Sonora, Baja California Sur, Jalisco, San Luis Potosí, Veracruz, Tamaulipas, Guanajuato and Michoacán.

The latter contributes with a planted surface of 448 ha of the national surface and with a yield of 11 095 t ha^-1 (^{SIAP, 2019}). However, one of the main species of chili cultivated in the states of Michoacán, Veracruz, State of Mexico and Puebla is C. pubescens (^{Barrios-Puente et al., 2014}), with great importance in national and foreign markets, and that its production has been increased by the high yields obtained, fruit quality and high profitability, the price fluctuates from $10.00 to $80.00 and is directly related to the season, size and fruit color.

Due to the above, there is a potential demand in growth for its consumption and production (^{Espinosa-Torres and Ramírez-Abarca, 2016}). This species is characterized by being mainly rainfed, where the most significant months for harvesting are from August to December (^{Barrios-Puente et al., 2014}). In Mexico, C. pubescens is commonly known as manzano chili, rocoto, canario or chili wax and in the state of Michoacán, it is recognized as perón chili ^{(Espinosa-Torres and Ramírez-Abarca, 2016}) and has wide genetic variability that can be used in genetic improvement programs as a result of the wide diversity of existing environments (^{Bobadilla-Larios et al., 2017}).

In this state, there are diverse homogeneous agroecological zones, which are generally favorable for the development of avocado cultivation (Gutierréz-Contreras et al., 2010) and some favor the development of perón chili, since good adaptation is observed in areas of shade where warm air mixes with cold wind, under the avocado trees (SAGARPA, 2013).

However, C. pubescens is known only under cultivation and the lack of knowledge about diversity and phylogenetic relationships limit its genetic improvement (^{Pérez et al., 2015}). In this sense, the related phytogenetic resources acquire great relevance because of the genetic potential they present and because they are the basis for obtaining improved varieties (^{Moreno-Pérez et al., 2011}).

According to the above, there is a need to know the distribution, variability and genetics of the materials retained by the producers in their plots (^{Martínez-Sánchez et al., 2010}) to avoid the loss due to genetic erosion as a consequence of disasters caused by factors adverse effects such as demographic changes, droughts, destruction of natural areas, substitution of landraces, pests and diseases, among others (^{CONABIO, 2011}; ^{Toledo-Aguilar et al., 2016}).

In this sense, it is necessary to rescue and characterize (^{IBPGR, 1983}; ^{UPOV, 2006}) the genetic diversity cultivated with the purpose of generating information on the variability of this species that prevails in different regions and agroecological zones of the state of Michoacán. It is also important to explore alternatives that allow us to know and conserve important characteristics to lay the foundations for the genetic improvement of the species.

In the present research work the collection of perón chili morphotypes in different agroecological zones of the state of Michoacán was proposed in order to evaluate the existing morphological variability, as well as the identification of important morphological characteristics to be used in breeding programs of the species.

Materials and methods

The research was carried out in the Phytogenetic Resources Laboratory (LAREFI) of the Advanced Research Unit in Agrobiotechnology (UIIA) of the ‘Presidente Juárez’ School of Agrobiology, located in the city of Uruapan, Michoacán, Mexico. The germplasm collection was carried out through field trips made from February 2016 to August 2017, due to the association of C. pubescens with avocado cultivation, mainly in the area known as the avocado producer fringe. Michoacán, located in the physiographic province of the transversal volcanic system between the parallels 18° 45’ and 20° 6’ north latitude and the meridians 101° 47’ and 103° 13’ west longitude, where ten homogeneous agro-ecological zones predominate (Gutierréz-Contreras et al., 2010).

In total, five fruits and five leaves of 31 cultivated varieties were taken at random. The characterization of fruit, leaf and seed was made based on the morphological descriptors proposed for Capsicum spp. (^{CATIE-IPGRI-AVRDC, 1995, UPOV, 2006}). In total, 27 variables were considered, six of leaf: color, shape, margin of leaf blade, pubescence, length and width; 16 of fruit: volume, presence of anthocyanin stains or stripes, color, shape, length, width, weight, length of the fruit pedicel, thickness of the wall, shape of the fruit in union with the pedicel, neck at the base of the fruit , shape of apex of fruit, appendix in the fruit, transverse wrinkling, number of locules and type of epidermis and five of seed: color, texture, size, weight and number of seeds per fruit.

The experimental design used was completely randomized with five repetitions. For the analysis of qualitative variables, double-state and multi-state data were given to which nominal values were assigned with logical sequence and quantitative variables were standardized to whole numbers with two decimals except for the number of locules and number of seeds per fruit. Subsequently, to determine the existence or not of statistical differences between the characters evaluated, a one-way analysis of variance was performed with the Proc Anova procedure and to evaluate significant statistical differences between cultivated varieties, the Tukey test was used (α≤ 0.05).

In order to select the variables that provided the most information to explain the total variance among the cultivated varieties, 24 variables were selected that presented significant statistical differences, on this list a discriminant analysis of Stepwise was practiced (^{Romano and Wolf, 2005}). Subsequently, to eliminate the correlated variables and avoid problems of collinearity in the matrices, a multiple correlation analysis was made (^{Pearson and Filon, 1898}).

In order to determine the similarity between the cultivated varieties, multivariate statistics tools were used, such as principal component analysis (PCA) and cluster analysis (AC) with the Ward grouping method, with the pseudo F test was determined on degree of reliability of the groups.

For the analyzes described, the SAS computation program version 9.2 was used (^{SAS, 2012}). Finally, to determine the possible relationship of the morphological characteristics with the origin of the cultivated varieties, the georeferencing information of the origin of the cultivated varieties was transformed (^{FCC, 2018}) to construct a matrix of geographical distances (DG), which was correlated with the matrix of Euclidean distances (DE) of morphological characteristics, by means of the Mantel test with the Genalex statistical package (^{Peakall and Smouse, 2012}).

Results and discussion

In four homogeneous agroecological zones (^{Gutiérrez-Contreras et al., 2010}) of the Michoacán avocado producer fringe, it was possible to identify perón chili, in total, 31 cultivated varieties of perón chili with contrasting morphological characteristics. The cultivated varieties were located between 1 623 and 2 282 meters above sea level, height ranges that coincide with those reported for the cultivation of this species (^{Yáñez et al., 2015}; ^{Espinosa-Torres and Ramírez-Abarca, 2016}; ^{Caballero-Gutiérrez et al., 2017}).

Perón chili is identified as a species that has genetic variability that can be used in breeding programs (^{Arias et al., 2017}). In this study, the cultivated varieties of perón chili showed significant statistical differences (α< 0.05) for 20 variables. In Table 1, simple statistics are presented for each of the variables analyzed. The coefficients of determination (R²) obtained ranged from R²= 0.29 (texture of the seed) to R²= 0.82 (fruit length), indicating that the data fit the model used for the analysis of the information, in addition, most of the variables presented high predictive capacity of the variance.

The standard deviation (SD) in most of the collections was low, except for the variables: fruit weight (SD= 8.17), fruit volume (SD= 9.12) and number of seeds per fruit (SD= 17.31). On the other hand, the variability of the evaluated characteristics and interpreted by the coefficient of variation (CV) showed that the measured characters have different levels of dispersion, the characters with the greater variation between collections were: appendix of fruit, vestige of flowering (CV= 95.30) and presence or absence of anthocyanin spots or streaks (CV= 91.68), in contrast to length (CV= 8.81) and fruit width (CV= 9.25).

Similar results have been obtained when characterizing populations of C. annuum, C. baccatum, C. chínense and C. frutescens (^{Palacios and García, 2009}). In this sense, when analyzing collections of C. annuum, C. frutescens and C. chinense CV greater than 25% were obtained for 75% of the evaluated descriptors (^{Villota-Ceron et al., 2012}), in collections of C. annum (timpinchili) the highest CV values were determined for stem length and crown diameter with a CV greater than 50% (^{Alonso et al., 2012}).

The wide variability of morphological characters presented by the genus Capsicum has been observed in different works (^{Pardey et al., 2009}). Although different species belonging to the genus Capsicum share traits in common, each species has its own characteristics, of importance in the selection processes (^{Palacios and García, 2009}). However, the variables evaluated in fruit are mainly considered to discriminate useful genetic variability in breeding programs (^{Pardey et al., 2006}).

For example, the thickness of pulp, weight, length and volume of fruit have been reported as important characteristics in the explanation of the variation that exists within the genus Capsicum (^{Medina et al., 2006}; ^{Castañon-Najera et al., 2008}; ^{Bozokalfa et al., 2009}; ^{Singh et al., 2017}). The results obtained from the stepwise discriminant analysis indicate that the variables that do not provide information on the variability between the cultivated varieties are the volume of fruit and type of epidermis.

Additionally, the results obtained from the stepwise discriminant analysis indicated that the variables: volume and thickness of the fruit wall, do not provide useful information to explain the variation between the cultivated varieties of manzano chili. Likewise, when reviewing the Pearson correlation values, these variables showed a high correlation with the weight and width of the fruit.

In Table 2, there are 27 significant correlations (r> 0.05) identified among the 27 variables evaluated in leaf, fruit and manzano chili seed. Of the 27 correlations, five correlations were identified with values of r≥0.8, the fruit weight with volume and width of fruit (r= 0.9521 and r= 0.8383, respectively), width of fruit, with volume of fruit (r= 0.8197), number of seeds with the weight of seeds (r= 0.8044) and width of the leaf with leaf length (r= 0.7841). 20 correlations with value r≤ 0.5.

In contrast, two correlations with negative value, length of fruit with shape of fruit apex (r= -0.3108) and fruit color with fruit width (r= -0.3114) were identified. Similarly, variables with high values of correlation between leaf and fruit characteristics in collections of the genus Capsicum have been identified (^{Villota-Ceron et al., 2012}).

In hybrids of C. pubescens the close association of the variables height of the plant, length and width of the leaf is demonstrated, in addition, the effect on the bearing and vigor of the plant as well as on the size of the fruit is pointed out (^{Arias et al., 2017}). The weight of the fruit is a dominant characteristic, related to the size of the fruit and the thickness of the pericarp; these characteristics are important due to consumer demand, since the latter is attracted by large and showy fruits, and fruits with thicker pericarp are more resistant in post-harvest handling (^{Lannes et al., 2007}).

However, the susceptibility to fruit rot caused by Phytophthora was correlated with the thickness of the pericarp; that is, chili with thicker pericarp tend to be more susceptible to P. capsici for this reason, small fruits are considered in breeding programs for their characteristics of resistance to insects and diseases (^{Naegele et al., 2016}) . The results obtained from the analysis of principal components (CP) indicated that the total variance explained, with the first three components was only 45.89%, while with nine, 80% is explained.

Similar results have been obtained in works of morphological characterization in situ in the genus Capsicum, since it has been observed that in this type of work it is not possible to explain more 80% of the variance with the first three components (^{Castañon-Najera et al., 2008}; ^{Martínez-Sánchez et al., 2010}; ^{Alonso et al., 2012}; ^{Villota-Ceron et al., 2012}; ^{de la Cruz-Lázaro et al., 2017}). Similar results in this type of work in other species have been reported (^{Medina-Torres et al., 2015}, ²⁰¹⁶; ^{Toledo-Aguilar et al., 2016}).

In the Figure 1 shows the distribution of 31 cultivated varieties from four homogenous agro-ecological zones based on three main components. When analyzing the eigenvectors with values above 0.3, it is determined that the CP1 is a function of the presence of the neck at the base of the fruit (CBF), leaf length (LH) and fruit length (LF), the CP2 is determined according to the weight of the seed (PS), fruit volume (VOL), width (ACHF) and fruit weight (FP), finally, CP3 is a function of fruit color (CF), transverse wrinkling (ARRT) ), number of locules (NL) and seed color (CS).

Figure 1 Dispersion of cultivated varieties of perón chili from four homogeneous agro-ecological zones of the state of Michoacán; through, of the first three components. Variables associated with CP1= neck in the fruit base (CBF); leaf length (LH) and fruit (LF); variables associated to CP2= seed weight (PS); volume (VOL); width (ACHF) and fruit weight (PF); variables associated with CP3= fruit color (CF); transverse wrinkling (ARRT); number of locules (NL); and seed color (CS).  

Similar results have been reported in others in other species, for example, in C. annuum 74% of the total variance is indicated, explained with the first three main components (^{Martínez-Sánchez et al., 2010}) as a function of days to flowering and fruiting, pericarp thickness, corolla length and width. The most important features to explain the variability between cultivated varieties of different species of the Capsicum genus are those related to the fruit and the architecture of the plant (^{Castañon-Najera et al., 2008}; ^{de la Cruz-Lázaro et al., 2017}), these results coincide with those obtained in the present investigation, since the width and length of leaf, shape, length, width and fruit weight have also been identified among the main morphological characteristics that favor the explanation of variability between collections.

In the dendogram generated from the cluster analysis, groups of cultivated varieties from the same locality were identified, as well as the association of varieties cultivated by the agroecological zone of the place of origin (Figure 2). In the germplasm collected there is a wide variability, these differences may be related to the places of origin. It has been identified that the temperature and the quantity of water available during the growth and reproduction of the plants are important factors for the differentiation of the populations of wild chili that grows in natural conditions (^{Hernández-Verdugo et al., 2012}).

Figure 2 Grouping of 31 cultivated varieties of C. pubescens from four homogeneous agroecological zones* of the state of Michoacán; Toreo el bajo (TOB); Tingambato (TIN); Tiamba (TIA); and Uruapan (URU).  

In contrast, the grouping of materials shows no relationship between species or collection environments (^{Villota-Ceron et al., 2012}). The results obtained on the variability and its relation with the origin of the cultivated varieties coincide partially with those indicated by the cited authors, in this work, the existence of groups of cultivated varieties clearly related to the agroecological zones was determined, although they were also identified groups without any apparent relationship to their place of origin. The results obtained in this study corroborate that the existing variability in C. pubescens is mainly explained by characteristics evaluated in the fruits (Table 3).

Finally, the results obtained from the Mantel test indicate that there were no significant correlations between the origin of the cultivated varieties and the morphological characteristics evaluated (Figure 3); nevertheless, the data used to correlate the variability with the origin are: latitude north and longitude west, without taking into account the agroecological characteristics of the sites of origin. Therefore, the variability identified among cultivated varieties of C. pubescens is not directly related to geographic location but may be influenced by homogeneous agroecological zones of provenance.

Figure 3 Mantel test, linear correlation between distribution of 31 cultivated varieties of C. pubescens based on Euclidean distances generated by morphological characteristics and geographical distribution according to the geographic coordinates of origin.  

In species of the Capsicum genus, the variability within and between species is not necessarily related to the zones of origin (^{Bobadilla-Larios et al., 2017}). Therefore, the grouping of the varieties cultivated in the present study may be influenced by gene flow since the appearance of new forms, new genotypes or hybrids is mainly due to the exchange of germplasm among farmers (^{Pérez et al., 2015}).

Conclusions

The cultivated varieties of perón chili from four homogeneous agro-ecological zones of the state of Michoacán presented a wide morphological variability in leaf tissue, fruit and seed. The characteristics evaluated in the fruit stood out as the most important variables to explain the total variation identified, in addition, that these characteristics can be used as parameters of fruit quality of perón chili.

On the other hand, it was not possible to establish any relationship between the morphological variability and the sites of origin of the cultivated varieties of perón chili evaluated. However, the idea that the morphological characteristics are influenced by the agro-ecological zones of the site of origin and by the gene flow due to the exchange of germplasm among farmers is not discarded.