Introduction

Salvia hispanica L. is an herbaceous plant of the Lamiaceae family, native to the mountainous areas of southwestern Mexico, Guatemala, and Nicaragua (^{Lobo et al., 2011}). The basis of the diet of the Indigenous peoples of Mexico, it was one of the four main crops of the Aztecs, surpassed only by corn (Zea mays L.), beans (Phaseolus vulgaris L.) and chili (Capsicum annuum L.). During the conquest of New Spain, its production decreased until almost disappearing, due to the reduction of the pre-Hispanic population and the implementation of cereal cultivation (^{Xingú et al., 2017}).

In recent decades it has resurfaced, the nutraceutical properties and its attractive nutritional benefits have expanded its consumption (^{Xingú et al., 2017}), it has oil with 68% of α-linolenic acid, the most important of the omega-3 fatty acids for human consumption, which makes it the richest vegetable source in antioxidants (^{Orona-Tamayo et al., 2017}), vitamins B1, B2 and B3 (^{Jamshidi et al., 2019}), fiber, proteins and minerals such as phosphorus, calcium, potassium, magnesium, iron, zinc and sodium (^{Michajluk et al., 2018}).

In addition, it has medicinal properties (^{Deka and Das, 2017}), has beneficial effects for the treatment of metabolic syndrome (^{Lombardo and Chicco, 2017}), regulates blood glucose and promotes blood clotting (^{Nieman et al., 2012}), decreases bad cholesterol and triglycerides, and improves intestinal function (^{Sandoval-Oliveros and Paredes-López, 2013}). The global demand for chia began in the nineties, it is currently grown in Argentina, Bolivia, Paraguay, Australia and Mexico mainly (^{Busilacchi et al., 2015}), where its consumption increases day by day, being exported to Peru, the United States of America, Chile, Germany, the Netherlands, the United Kingdom, Denmark, Japan, Canada, New Zealand, Singapore and South Africa (^{Suárez, 2018}).

In Mexico, the commercial production of this species takes place in eight states. Although during the 2006-2009 period, the area sown with chia was less than 50 ha, by 2010 it increased to 2 300 ha, a figure that increased constantly until the 2013 cultivation cycle, which exceeded 18 000 ha. However, from 2014, it registered a decreasing trend and during the 2017 spring-summer cycle, it was grown only on 5 400 ha with a production of 3 200 tons (^{SIAP, 2019}). Reduction due to factors such as: lack of experience on the cultivation, ignorance of nutritional properties and limited information on improved varieties (^{Sosa-Baldivia and Ruiz-Ibarra, 2016}).

The state of Jalisco is the largest producer, concentrating more than 65% of the cultivated area which contributes a volume of more than 2 000 tons, equivalent to 63% of production (^{SIAP, 2019}). Chia cultivation is mostly based on regional genotypes. Although germplasm banks have been established in institutions such as the Chapingo Autonomous University (UACH, for its acronym in Spanish), the National Institute of Nuclear Research (ININ, for its acronym in Spanish), the Institute of Agricultural, Aquaculture and Forestry Research and Training of the State of Mexico (ICAMEX, for its acronym in Spanish), the collection of accessions of the company Chíablanca SC de RL located in Acatic, Jalisco stands out, which concentrates collections from the different producing areas.

Information on the characterization of materials of this species is scarce, wild, semi-domesticated and domesticated populations have been typified (^{Calderón-Ruiz et al., 2021}). ^{Hernández and Miranda (2008}) studied three ecotypes of cultivated chia, finding similarity in seed size and inflorescence density, but with differences in biological cycle, length and width of corolla, width of inflorescence and height of plant, and they concluded that among the morphological structures that differentiate cultivated S. hispanica from the wild are: flower size, density of whorls in the inflorescence, seed weight and duration of the biological cycle.

Studies carried out by ^{Sosa-Baldivia et al. (2017}) report potential yields of 1 723 kg ha^-1, which they relate to the number of plants per m², plant height and main inflorescence length, while ^{Grimes et al. (2018}) reported a production of 1 274.7 kg ha^-1 of the Sahi Alba 914 variety. Yields are related to a higher number of branches plant^-1 and inflorescences plant^-1 (^{Pereira et al., 2020}). Currently, studies have focused on demonstrating the properties as a functional food (^{Grancieri et al., 2019}) but the work of describing the available materials of this species has been insufficient, so the present research arose with the aim of agronomically characterizing chia accessions from the main producing regions of Mexico.

Materials and methods

Results and discussion

Significance (p≤ 0.01) was obtained between environments (E), between chia accessions (C) and in the interaction of accessions (C) by environments (E) for dry weight of plant, stem diameter, plant height, number of fruits per spike, spike length, spike length from node, number of branches per plant, number of spikes per plant, grain weight per plant, harvest index and yield in kg ha^-1 (Table 3).

Table 4 shows that, although RSL surpassed the rest of the localities in plant height, RSL1 showed the highest dry weight per plant, grain weight per plant and yield. RSL, RSL1, SFT and XAL have statistical similarity in spike length from node and number of fruits per spike. RSL, RSL1 and SFT surpass the rest of the localities in number of spikes per plant. RSL and RSL1 have greater spike length and number of branches per plant. This allowed establishing that the best behavior occurred in the RSL and RSL1 localities, which profiles this site as a potential area to promote the development of the crop of chia. The differences in the productive parameters of the crop of chia in the evaluated localities can be attributed to the fact that the conditions of each locality can influence the development and production of the collections (^{Durán et al., 2016}), although the genotype effect is the most marked (^{Busilacchi et al., 2013}).

In relation to accessions, in Table 5, selections 1 (black seeds) and 12 (white seed) stand out in grain weight per plant and yield (exceeding 1 400 kg ha^-1), the weight of seed influences the yield, the marbled gray and white seeds are the heaviest compared to those of uniform brown color (^{Rovati et al., 2012}). Materials 19 and 23 showed seed weight per plant and yields of less than 50 kg ha^-1, genotypes affected by frosts in full flowering, it would be convenient to evaluate them in areas with less risk of frosts or modify the sowing season because the plant is sensitive to low temperatures (^{González, 2016}).

Accessions that exceed yields of 1 000 kg ha^-1 of seed also exceed 30 flower spikes, 90 cm in height and 34 g of dry weight of plant, it could be deduced that these variables are closely related to seed yield (^{Karim et al., 2016}).

The dendrogram shows that, at a Euclidean distance of 200, five groups formed (Figure 1). Set 1 was formed by accessions 17 and 18, which presented physiological maturation at 160 days after sowing, unlike those of intermediate cycle which have their production at 150 days. Production is low, 306 kg ha^-1 (18) and 502 kg ha^-1 (17), but they are within the yields of the national average of 500 kg ha^-1 (^{SIAP, 2019}).

Cluster two consisted of accessions 19 and 23, which showed flowering at 160 days, but showed cold damage in the frost season, so their production was minimal, of 20.8 kg ha^-1 (19) and 45.7 kg ha^-1 (23), this confirms that temperatures below 5 °C affect the crop of chia (^{Baginsky et al., 2016}) and that, in temperate climates, more biomass accumulates and they produce less seed, contrary to when they grow in warm environments, in these conditions they accelerate the reproductive phase and produce more seed (^{Medina-Santos et al., 2019}).

Figure 1 Dendrogram from 12 agronomic variables of 32 chia collections. 

Accessions 15, 20 and 25 form a subgroup of cluster three, which can be considered of early cycle, since their flowering occurred at 90 days and their physiological maturity at 120 days, they share the average presence of 15 branches per plant, this subgroup integrates accessions 29 and 3 of intermediate cycle (150 days), with 16 to 18 branches and 47 to 48 flower spikes per plant. With similar seed production, another subgroup consists of accessions 2 (1 043 kg ha^-1), 22 (1 027 kg ha^-1), 5 (1 037 kg ha^-1), 16 (1 039 kg ha^-1) and 21 (1 053 kg ha^-1). Accessions 7 and 14 share the same production 1 071 kg ha^-1 and 1 064 kg ha^-1 respectively, the grain weight per plant 7 (6.54) and 14 (6.52) and number of flower spikes per plant 7 (33) and 14 (35).

Group four was formed by accessions six (1 141 kg ha^-1) and 28 (1 115 kg ha^-1), which form a subgroup sharing the same grain production and weight per plant, another subgroup is formed by collections 8 and 22 with yields of 1 175 and 1 188 kg ha^-1 respectively, as well as the same grain weight per plant. The subgroup formed by accessions 10 and 26 share stem diameter (7.2 mm), spike length from node, seed production per plant and yield. Another subgroup formed by accessions 13 and 30 has the same seed production per plant (4.7 g), yield per hectare of 777-781 kg and number of flowers per floret in spike. Similar yields of the accessions of this group have been obtained in Petacal, Jalisco with local cultivars (^{Sosa-Baldivia et al., 2017}).

Cluster five was integrated by accessions 1 and 12, which were the ones that had the highest seed weight per plant and production, with yields exceeding 1 400 kg of seed per ha. Fruit or seed production and yield are variables that allow determining the ideal genotypes to implement in the search for cultivation areas (^{Bochicchio et al., 2015}).

Principal component analysis

The first four principal components explain 74% of the agronomic variability of 32 accessions of S. hispanica. ^{Sánchez (1995}) mentions that this percentage is reliable to properly interpret the correlations that exist between them. The first component with 30.43% was related to yield. The second component with 21.39% was defined by the variable of seed weight per plant, the third principal component 12.25%, defined by the number of spikes per plant, and the fourth component with 10.24 of the variability generated by spike length, they collected the variation not gathered by the first, presenting the highest factorial coefficients.

In principal component analysis, the new factors (or components) are independent of each other, that is, a variable must have high coefficients with only one factor and there should be no factors with similar coefficients (^{Restrepo et al., 2012}). The variables yield, grain weight per plant, number of fruits per spike, number of spikes per plant and spike length have a positive and significant contribution, which allows specifying the contribution of the variables to the principal components and their relationship with the explained variation (Figure 2). The variables studied tend to be grouped, with an acceptable degree of agreement in their location within the quadrants (^{Olivares and Hernández, 2020}).

Figure 2 Multidimensional representation of the two principal components of 32 collections of S. hispanica.  

Conclusions

The characterization showed that there is a wide agronomic variability between the accessions evaluated, the factors yield, number of spikes, seed weight per plant and plant height are variables that allow the best materials to be selected; selections 1, 2, 12 and 22 have outstanding agronomic characteristics, which can be established with potential yields attractive to producers in the study area, sowing in a timely manner when the rainy season begins and thus avoid frost damage or as a basis for developing genetic improvement programs for S. hispanica L.