Introduction

Mexico ranks as the world’s leading producer of avocado, a place that obtains the 149 185 hectares planted, the state of Michoacán, is placed as the leading producer of this fruit with 195 042 t representing 94.9% of national production (^{SIAP, 2019}). In Michoacán, avocado cultivation is established in a region known as the ‘avocado producer fringe of the state of Michoacán’ (^{Gutiérrez-Contreras et al., 2010}). There is a wide genetic diversity of P. americana Miller variety Drymifolia it has been used as rootstock of the Hass variety (^{Cuiris-Pérez et al., 2009}).

For being a source of genes for resistance to physical factors, pests and pathogens (^{Rincón-Hernández et al., 2011}); however, the enormous phytogenetic wealth is threatened by the destruction of ecosystems and substitution of traditional cultivars by improved cultivars (^{Lorea-Hernández, 2002}). Consequently, during the last years, efforts have been added for the conservation and characterization of these genetic resources (^{Gutiérrez-Diez et al., 2009}).

In particular, in the state of Michoacán, work has been done on the characterization of the creole avocado germplasm; these works include genotypic information (^{Cuiris-Pérez et al., 2009}) and phenotypic information (^{Guillén-Andrade et al., 2013}). The importance of studying the chemistry of plants is that they have developed various defense strategies under conditions of biotic and abiotic stress; for example, the plant synthesizes secondary metabolites (MS) that cause plants to be unpleasant or toxic to some herbivores (^{Anaya-Lang and Espinoza-García, 2006}).

Some with larvicidal biological activity (^{Senthilkumar et al., 2008}), bactericidal and fungicidal insecticide (^{Khokra et al., 2008}; ^{Moreno et al., 2009}; ^{Hanamanthagouda et al., 2010}), in addition to their ecological importance by participating in the plant adaptation processes, attraction of pollinating insects and seed dispersers (^{Sepúlveda-Jiménez et al., 2003}).

The variation of the chemical profiles of the Persea genus has been documented in several studies (^{Quing-Yi et al., 2009}; ^{Torres-Gurrola et al., 2009}; ^{Rincón-Hernández et al., 2011}) in the most (^{Torres- Gurrola et al., 2016}) reported a total of 363 identified MS, corresponding to 13 chemical groups, of the total, 258 are related in different tissues of Persea americana: 125 have been reported in pericarp, 109 in leaf tissue, 17 in seed, three in bark, two in flower and two in root. Due to the importance of knowing the chemical profiles, as defense mechanisms, of the genetic resources of P. americana Miller variety Drymifolia.

With the objective of knowing the foliar chemical variability of the 54 collections of this species, established in the germplasm bank of the Faculty of Agrobiology, dependent on the Michoacán University of San Nicolás de Hidalgo, as an integral part of a strategy to evaluate its potential as rootstocks with characteristics of resistance to pests and diseases, among other attributes additionally, the need to establish the possible relationship of the MS present in the collections with their geographical origin was raised.

Materials and methods

The research was developed in the Plant Genetic Resources Laboratory (LAREFI) of the Advanced Research Unit in Agrobiotechnology (UIAA) of the ‘Presidente Juárez’ Faculty of Agrobiology, of the Michoacán University of San Nicolás de Hidalgo, located in the city of Uruapan, Michoacán, Mexico, at the coordinates 19° 25’ 10’’ north latitude and 102° 03’ 30’’ west longitude at 1 620 meters above sea level. The process of extraction, identification and quantification of MS was carried out at the Center for Research in Ecosystems (CIEco) of the National Autonomous University of Mexico-Campus Morelia.

The genetic material analyzed consisted of 54 accessions of P. americana Miller variety Drymifolia, with four individuals per accession, the accessions come from localities belonging to the municipalities of Ario, Los Reyes, Salvador Escalante, Tacambaro, Tingambato, Uruapan and Ziracuaretiro, all of them located in the Michoacán avocado strip between coordinates 18° 45’ and 20° 6’ north latitude and 101° 47’ and 103° 13’ west longitude (^{Guillén-Andrade et al., 2007}; ^{Gutiérrez-Contreras et al., 2010}).

The chemical profile of each of the accessions was determined from leaf tissue of mature leaves and with the use of a gas chromatograph (Agilent HP6890^®), provided with a mass detector (Agilent 6890^®). The procedures for extracting extracts, the conditions of the gas chromatograph for injecting and analyzing the samples, the identification of MS based on the information contained in the mass spectrum library NIST 05 (National Institute of Standars and Technology) and the Quantification of MS were made based on what was described by ^{Torres-Gurrola et al. (2009)}.

The content and concentration database (mg g^-1 dry leaf) of each of the identified MS was subjected to a one-way analysis of variance, based on a completely randomized experimental design, with four repetitions, considering to each accession as a treatment. Each observation was represented in the statistical model corresponding to the experimental design, by means of the following expression: Yij= M +Ci + eij. Where: Yij= concentration of MS in collection i and repetition j; M= general mean of the experiment; Ci= effect of the collection i; Eij= experimental error in collection i and repetition j; i= 1, 2, 3, ..., 54 and j= 1, 2, 3, 4.

This analysis was done using the Proc Anova procedure of the statistical package SAS⁽ version 9.4 (^{SAS Institute Inc., 2012}). Based on the analysis of variance, the MS were selected for which there were statistically significant differences (p< 0.05). Later, on the list of those selected MS, a discriminant analysis of Stepwise was carried out (^{Romano and Wolf, 2005}) to debug and maintain only that MS that contributed more information to the total variance.

Additionally, a correlation analysis of ^{Pearson and Filom (1898)} was carried out to eliminate highly correlated variables, and to avoid problems of collinearity in the matrices. Subsequently, a matrix of averages was obtained to carry out a principal components analysis (PCA), using the matrix of correlations; the Euclidean distance matrix (DE) between accessions was calculated and with it, the cluster analysis was done by means of the Genalex statistical package (^{Peakall and Smouse, 2006}).

Finally, the corresponding dendogram was constructed using the Neighbor-Joinning grouping method. All analyzes were made with the statistical package SAS⁽ version 9.4 (^{SAS Institute Inc. 2012}). The geographical coordinates of the accession source sites were converted to decimals with the Federal Communications Commission program (http: // transition.fcc.gov/mb/audio/bickel/DDDMMSS-decimal.html), with this data was used to construct a Geographical Distances (DG) matrix, which was correlated with a matrix of Euclidean distances, using the Mantel test with the GenAlex statistical package (^{Peakall and Smouse, 2006}; ^{Flanagan, 2006}).

Results and discussion

In the analysis of 54 accessions of established in the germplasm bank of the Faculty of Agrobiology, the results obtained indicate the presence of 47 MS of the total, there is a first group of 31 that have been reported in P. americana Miller and related species: α-pinene, sabinene, β-pinene, eucalyptus, β-cis-ocimene, estragole, caryophyllene oxide, limonene, α-cubebene, phytol, heptacosane, squalene, chavicol, β-cubebene, caryophyllene, α-humulene, germacrene D-4-ol, tetradecanoic acid, copaene, cubenol, decane, phenyl ethyl alcohol, methylenegolol, nerolidol (E), tetradecanal, (-elemene, ethyl linolenate, hexadecanoic acid, methyl hexadecanoate, myrcene and oleic acid (^{Quing-Yi et al., 2009}; ^{Torres-Gurrola et al., 2009}; ^{Rincón-Hernández et al., 2011}; ^{Torres-Gurrola et al., 2016}).

A second group, consisting of 11 MS, have not been reported for P. americana and related species; however, for these there are reports of their presence in other plant species. 4,8,13-Duvatrien-1,3-diol has been identified in tobacco as growth inhibitor 8,11,14 eicosatrienoic acid (z,z,z)- in Melittis melissophyllum (^{Velasco-Negueruela et al., 2004}), eugenol acetate, has been reported as the main component in the essential oil of Syzigium aromaticum and with antioxidant activity; cis, cis, 7,10-hexadecadienal in Euphorbia heterophylla with cytotoxic activity, antioxidant and antimicrobial cis,cis,cis-7,10,13-hexadecatrienal in Azadirachta indica and Allamanda cathartica with larvicidal activity.

The elemicin in Myristica fragrans and Daucus carota with toxic and antibacterial activity and in Syzigium aromaticum with antioxidant activity palmite in Annona diversifolia acting as an anticonvulsant (^{Cano-Europa et al., 2010}), α-tocopherol in Elaeis oleifera and Vaccinium meridionale with antioxidant activity cycloartenol acetate in propolis with antimicrobial activity 2H-pyrene, 2- (7-heptadecinyloxy) tetrahydro- me was identified in the medicinal plant Andrographis paniculata 2-methilenecholestan-3-ol, was reported as a component of the essential oil of the flowers of Artemisia austro-yunnanensis with antioxidant activity (^{Chen-xing et al., 2014}) and in Alstonia scholaris: plant used in herbal medicine for its medicinal properties (^{Islam et al., 2013}).

Finally, a third group of five MS were identified for which there are no reports of their presence in plant species: undecane and tridecane have been reported as defensive compounds in Loxa deducta and Pellaea stictica methyl arachidonate as a growth factor in animals. α-glyceryl linolenate, is used as an industrial solvent, undecane 4-methyl has been identified in wastewater and crude oil.

Based on the works, which have been published on this topic, in this research a total of 16 new metabolites were present in foliar tissue of P. American Miller Drymifolia variety and belonging to nine chemical groups: esters (4), alcohols (2), phenolic compounds (1) acids; (1) ether; (1) alkanes; (3) aldehydes; (2) acetone; (1); and polyterpenes (1). In Table 1, the name of the metabolite and its chemical structure are presented.

When analyzing the concentration and frequency at which the 47 MS were identified in the 54 accessions, it was determined that β-pinene, estragole, caryophyllene, hexadecanoic acid, heptacosane and α-tocopherol were identified in all the accessions and, of these last, estragole stood out for presenting the highest percentage values (26.5327 mg g^-1 of concentration). In contrast, chavicol and eugenol acetate were present in only five of the 54 accessions analyzed. On the other hand, phenyl ethyl alcohol was the one that showed the lowest percentage (0.0164 mg g^-1 of leaf) concentration in the collections analyzed. In Table 2, the secondary metabolite information, retention time, Kovats index, concentration (%) and frequency of each of the MS identified in the 54 accessions are presented.

Estragole was the secondary metabolite highlighted by being present in higher concentration and in the 54 collections analyzed. These results are similar to those reported in similar works in germplasm different Persea species from different regions of Mexico (^{Rincón-Hernández et al., 2011}). The evaluation of the content of this MS in each of the collections is interesting, for example, the interaction avocado creole-Trioza anceps has been evaluated, the trees with the highest concentration of estragole showed less presence of leaf galls (^{Torres-Gurrola et al., 2009}).

In addition, estragole has been identified as the main component in other species and, with important biological defense activities, as an insecticide (^{López et al., 2008}) in Ocimum basilicum induced contact mortality to the rice weevil (Sitophilus oryzae) (Pascual-Villalobos et al., 2004), it has also been used against the red flour beetle (Tribolium castaneum Herbst) and the grain borer (Rhyzopertha dominica) in Pimpinella anisum, it was reported with bactericidal effect, it has been related to antifungal activity (^{Fontenelle et al., 2008}), larvicide (^{Senthilkumar et al., 2008}) and genotoxic (^{Zani et al., 1991}).

In the present study, three previous collections of the municipality of Tingambato, Michoacan, Mexico, were those that presented the highest concentration of estragole, so it would be expected that these collections could present better defense attributes to be used as selected rootstocks.

In the Figure 1 shows the variation determined in the content of estragole in the 54 collections analyzed.

Figure 1 Histogram for concentration of estragole in 54 accessions of Persea americana Miller variety Drymifolia collected in the municipalities of Ario (ARI), Los Reyes (LES), Salvador Escalante (SES), Tacámbaro (TAC), Tingambato (TIN), Uruapan (URU) and Ziracuaretiro (ZIR) established in the germplasm bank of the Faculty of Agrobiology.  

The values of the coefficient of determination (R²) oscillated between 0.33 and 0.46, in relation to the coefficients of variation (CV), the values fluctuated between 98.13 and 896.43. The highest CV value was obtained for the eugenol acetate in contrast to the Cis, cis, cis 7, 10, 13-hexadecatrienal. Likewise, the average values ranged from 0.00108 to 0.28877 in percentage of total concentration.

Later, in the Stepwise discriminant analysis, the tridecane was identified as one of the MS that did not provide information to explain the variation in MS concentration and when reviewing the Pearson correlation values, the dean was highly correlated (<0.0001) with the tridecane.

These results are explained by the fact that the collections present differences in their gene expression and the chemical profile is subject to strong genetic control, as it is in other tree species (^{Langenheim and Stubblebine 1983}; ^{Gershenzon et al., 2000}; ^{McConkey et al., 2000}).

The results obtained from the one-way analysis of variance, performed for 47 MS content identified by analyzing 216 individuals from 54 collections of P. americana Miller variety Drymifolia, indicated significant statistical differences (p< 0.05) for concentration for ten of them and differences highly significant (p< 0.01) for phenyl ethyl alcohol and 8,11,14 -eicosatrienoic acid (z,z,z)-. Table 3 shows the simple statistics obtained for the 12 secondary metabolites that presented significant statistical differences.

To reduce the number of variables to be considered for the explanation of the variability in relation to the concentration of secondary metabolites, a principal component analysis (PCA) was carried out. The results indicated that the first three main components (CP), explain together 29.64% of the variance that exists between the accessions, while with 17 it is explained 80%.

Similar results have been obtained in works of in situ morphological characterization of Persea and other fruit species, since it has been observed that in this type of work is not achieved more than 80% of the variance explained with the first three components (^{Tofiño et al., 2012}; ^{López-Gúzman et al., 2012}; ^{Guillén-Andrade et al., 2013}; ^{Medina-Torres et al., 2015}; ^{Montes-Hernández et al., 2017}).

In Figure 2, the distribution of the 54 collections is shown according to three main components. When analyzing the eigenvectors with values above 0.3, it was determined that the CP1 is a function of the presence of the eucalyptus, caryophyllene and α-humulene, the CP2 was determined in terms of the decane, undecane, 4-methyl and tridecan. finally, CP3 is a function of chavicol and τ-elemeno. It is important to note that the tridecane and the undecane, 4-methyl with the highest weight in CP2 are reported for the first time in Persea americana Miller variety Drymifolia.

Figure 2 Dispersion of 54 collections of P. americana Miller variety Drymifolia from the germplasm bank of the Faculty of Agrobiology, based on three main components.  

On the other hand, the dendrogram presented in Figure 3, generated from the cluster analysis based on Ward’s grouping method, showed that accessions are not associated in a logical way in relation to their geographical origin. These results were corroborated when performing the Mantel test, the correlation value obtained was r= -0.064 and a significance of 0.069, which confirmed the non-relationship between the geographic origin of the collections and the MS content identified in 54 collections.

Figure 3 Complete clustering dendrogram of 54 collections of Persea americana Miller variety Drymifolia, based on data on the concentration of 47 secondary metabolites. The colored circles indicate the place of origin of the germplasm.  

These results may be masked because the leaf tissue samples were isolated from the collections established in the germplasm bank; that is, they are in a single environment.

The results of the aforementioned test are presented in Figure 4.

Figure 4 Linear correlation between the content of secondary metabolites and geographic distribution of 54 collections of collections of Persea americana Miller variety Drymifolia.  

Conclusions

The 54 collections established in the germplasm bank of the Faculty of Agrobiology of Persea americana Miller variety Drymifolia, are contrasting in the type and concentration of 47 identified secondary metabolites. Estragole was a constant in terms of content and greater concentration in all the germplasm analyzed. The identification of 16 new secondary metabolites in the evaluated germplasm is an important contribution of the present work.

Finally, no relationship was determined between the chemical profile and the geographical origin of origin of the collections analyzed.