Genetic Divergence Among Safflower Genotypes (Carthamus tinctorius L.) by Multivariate Analyzes

Carthamus tinctorius L. is an oil seed, used both for human consumption and for industrial purposes. It is a crop that presents wide adaptability to various ecophysiological conditions, although it presents great productive potential and wide adaptability, it is still necessary to obtain technical information regarding its cultivation and of cultivars adapted and improved. In this sense, the estimation of genetic divergence using multivariate techniques has become a common tool among breeders. In view of the above, this research aimed to evaluate the genetic divergence of safflower genotypes originate from the Germplasm Active Bank (BAG) of the Instituto Mato-grossense do Algodão (IMA-MT) by means of multivariate analysis, aiming at the extension of information of this culture. The genetic divergence was estimated using multivariate analysis based on the Euclidean average distance, using the clustering optimization methods of Tocher and Hierarchical “UPGMA”. The results obtained allowed to identify the existence of genetic divergence among the evaluated genotypes, highlighting genotypes 5 and 38, which presented greater genetic divergence, constituting in potential sources of interest for the use in program of genetic improvement that aim at the development of superior cultivars of safflower.


Introduction
Carthamus tinctorius L. is a member of the family Asteraceae, is an oilseed, used in both human and industrial purposes. It is a crop that presents a wide adaptability to several ecophysiological conditions, developing satisfactorily under low water availability in low fertility soils and in locations with temperature variations, thus being an alternative crop for Brazilian arid and semi-arid regions (Moura et al., 2015).
Although it presents great productive potential mainly due to the value of its oil and its wide adaptability, it is still scarce technical information about its cultivation and of cultivars adapted and improved in the Brazilian regions (Gerhardt, 2014). In this perspective, genetic improvement may help in this process, since one of the objectives of breeding is to increase the economic value of the species, increasing productivity, resistance to diseases and nutritional quality (Borém & Miranda, 2005).
For a breeding program to be successful there is a need for genetic divergence in populations that will be subjected to selection, that is, genetic variability in the population (Ivoglo et al., 2008). The genetic diversity is evaluated with the objective of identifying the hybrid combinations with greater heterotrophic effect, to identify the parents who, when crossed, allow the appearance of superior genotypes. In addition, the study of genetic divergence is important for the monitoring of germplasm banks, in addition to facilitating the knowledge of the genetic basis of the population (Ferrão et al., 2002;Cruz et al., 2004).
Among the predictive methods used in estimates of genetic distance is the average Euclidean distance, where it can be estimated by taking as a base data without repetitions (Carvalho et al., 2003). The estimation of genetic divergence using multivariate techniques has become common among safflower breeders (Shivani et al., 2010;Safavi et al., 2012;Zoz, 2015;Pavithra et al., 2015;Atole et al., 2018).
In this sense, the present research aimed to evaluate the genetic divergence of safflower genotypes by means of multivariate analysis, aiming at the extension of information of this culture for safflower breeding programs.

Site Localition and Characterization
The experiment was carried out in the experimental area belonging to the Empresa Mato-Grossense de Pesquisa, Assistência e Extensão Rural (EMPAER), in the county of Cáceres, Mato Grosso state, located at latitude 16°43'42'' South and longitude 57°40′51″ West with altitude of 118 meters, at BR 070, 12 km from Cáceres. The typical climate of the region, according to the classification of Köppen, is tropical, hot, humid and dry winter (Awa), with a period of rainfall ranging from October to April and from May to September (Dallacort et al., 2014). The soil is classified as Chernossolic Eutrophic Yellow Red Argissolo, with a medium clay texture (Arantes et al., 2012).
Were evalueted 50 safflower genotypes from the North American germplasm bank Western Regional Plant Introduction Station (WRPIS), obtained through the Germoplasm Resource Information Nertwork (GRIN), imported by Instituto Mato-Grossente de Algodão (IMAT-MT) and ceded to the Laboratory of Genetic Resources and Biotechnology (LRG&B) of Universidade do Estado de Mato Grosso (UNEMAT), university campus of Cáceres (Table 1).

Morphoagronomic Characterization
The preparation of the experimental area was carried out in the conventional system, performing manual sowing on May 1st, 2018, with fertilization of 450 Kg ha -1 of formulated (N P 2 O 5 K 2 O) 4-14-8 according to culture recommendations. The plots of each genotype consisted of four lines with 1 meter of length, with spacing of 0.50 centimeters between rows and 0.10 centimeters between plants, having a total of 10 plants per line. Basic management measures, such as manual weeding and irrigation, have been adopted by sprinkling whenever necessary. The

Statistical Analyses
Diversity among the 25 safflower genotypes genotypes for 12 traits was assessed by estimating Euclidean average distance. Based on this matrix, we used the Tocher optimization grouping methods and Hierarchical Method of Intermediate Cluster between Groups (UPGMA) to build the genetic distance between the genotypes in clusters. The criterion of Singh (1981) was also used to quantify the relative contribution of the characteristics to genetic divergence. All analyzes were performed using the computational resources of the Genes software (Cruz, 2013). Table 2 shows the descriptive analysis of the twelve quantitative traits evaluated in the 50 safflower genotypes, where it is possible to observe that the genotypes showed an average flowering time of 84.5 days, with the genotypes PI248852 and PI250083 showing a shorter flowering time 75 days and the later flowering genotype PI248620 at 104 days. These results are superior to those obtained by Shinwari et al. (2014), in Islamabad-Pakistan, evaluating 122 genotypes collected from several geographic ecosystems in the world, where the average for this parameter was 175.2 days, with minimum and maximum values of 160 and 188 days, respectively.

Results and Discussion
In relation to the crop cycle, safflower genotypes presented values ranging from 118 to 156 days, and genotypes PI343783, PI193473, PI250083 and PI369849 showed to be early, while genotypes PI248620 and PI305209 were found to be late (Table 2). These values are similar to those of Pavithra et al. (2015), in which they obtained values from 128 to 148 days, evaluating 150 safflower accesses in the Karnataka region of India during the years 2011/12. According to Galant et al. (2015), the safflower cycle tends to vary from 130 to 150 days, and may be an option for growing in the dry season, or for the dry period in some Brazilian agricultural regions.
For the number of ramifications per plant, the average of the evaluated genotypes was 13.58, and the genotypes PI248385, PI253899 and PI305198 presented values higher than 20 ramifications per plant (Table 2). Although this quantitative factor is important in the question of productivity, since the trend would be that the larger the number of branches the greater the number of inflorescences and consequently the greater the number of chapters, these characteristics should be carefully analyzed, since for the safflower crop, there is no interest in obtaining stalks with many ramifications due to differences in flowering rates within the chapter. Bellé et al. (2012), considers that many ramifications result in lack of uniformity in the anthesis, which reduces the quality of the stems.  Regarding plant height, the safflower genotypes evaluated obtained an average of 84.53 cm, ranging from 62 to 111 cm (Table 2). These results are similar to Silva (2013), where plant height varied between 57 and 136 cm. According to Gracia et al. (2010), the observed values, for the most part, are within the limits commonly observed in the improved safflower cultivars, whose averages normally ranging between 90 and 150 cm.
As for the number of seeds per chapter, the average obtained by the genotypes evaluated in the present study was 26.8 (Table 2). Shinwari et al. (2014) found similar results, in a work conducted in Islamabad-Pakistan, where this variable presented an average of 28.2.
Considering the number of chapters per plant, the genotypes evaluated presented values ranging from 6 to 59 (Table 2). In a study carried out by Silva (2013), in Botucatu-SP, evaluating 170 accessions of safflower, there was a variation from 9 to 78. This parameter of production is relevant, since the larger the number of chapters per plant, the greater the number of seeds produced, which can promote productivity increases. Hajghami et al. (2009), emphasizes that to obtain promising cultivars in breeding programs, should select the materials with the highest number of chapters per plant to obtain highly productive plants. Therefore, the genotypes PI195895, PI250203, PI250840, PI253899 and PI262447 are distinguished for this characteristic producing more than 50 chapters per plant.
For the characteristic diameter of the chapter the safflower breeding programs search genotypes that present larger diameter, since the greater the chapter the greater the capacity of flower formation and, consequently, the greater the number of seeds, thus favoring productivity (Silva, 2013). Among the genotypes evaluated in the present study, PI250840 presented the highest result with a diameter of 3.16 cm (Table 2). This value is higher than those obtained by Atole et al. (2018), evaluating 155 safflower genotypes evaluated in Maharashtra-India, whose maximum value was 2.78 cm and by Silva et al. (2015), evaluating 20 genotypes evaluated in Botucatu-São Paulo, the highest value was 2.4 cm.
The diameter of the stalk is a very important feature in a plant of the same safflower family, as the sunflower (Asteraceae). Considering that it allows less bedding of the crop, facilitating its management, treatments and harvesting (Biscaro et al., 2008). As previously reported by Anicésio (2014), the study of this variable also applies to safflower, because this crop presents bedding problems that hamper, among other steps, mechanized harvesting. Thus, in the present research the genotypes PI 195895, PI248620, PI253899, PI262443, PI279344 and PI305207 are outstanding, since they have a diameter with values equal to or greater than 12 cm.
For the weight of 100 seeds, the genotypes varied from 3.76 to 7.10 g and average of 5.05 g ( Table 2). These results are higher than those obtained by Pushpavalliet et al. (2017)  Regarding the variable plant yield, the average of the 50 safflower genotypes evaluated in the present study was 26.23 g, with a maximum value for genotypes PI305207 with a value of 70.28 g (Table 2). This result is similar to that obtained by the hybrid combination between the genotypes PI537697 and PI653152 in Olivo (2017), in which the grain yield per plant was 73.80 g.
The genetic divergence of the evaluated genotypes was based on the average Euclidean distance, in which the most dissimilar pair was composed of genotypes PI248620 and PI343783, this dissimilarity between these genotypes may be linked to their geographical origin, considering that genotype PI248620 is from Pakistan and genotype PI343783 is of Iranian origin (Table 1). These results seem to be a trend, as they were previously reported in studies conducted by Derakhsan et al. (2014), in which the genetic divergence of 42 genotypes of six species of Carthamus tinctorius L. was evaluated, via microsatellite markers, where the results indicated that, in most cases, safflower genotypes are divided into subgroups consistent with the country of origin, that is, genotypes of different geographical origin are expected to be divergent.
From the point of view of genetic improvement, the divergence between these genotypes is of great importance, since as pointed out by Cruz et al. (2004), it is recommended to cross between divergent materials, for maximum heterosis in the progenies increasing the possibility of genetic gains in the segregating populations.
In relation to similarity, genotypes PI262450 and PI305198 are the closest, this fact, can be considered, because of their origins, since both are from India, because they present this similarity, the crossing of this combination is not recommended, having since for genetic breeding programs variability is indispensable (Santos et al., 2014).

Figure 1
Note.   In a study conducted in the state of Maharashtra, India, evaluating 155 genotypes by Atole et al. (2018), plant height and number of seeds per chapters also stood out for discrimination of genetic diversity with 22.75 and 20.68%, respectively. In the research carried out by Tayade et al. (2015), in Akola, India, evaluating 155 genotypes and five safflower varieties, the number of chapters per plant characteristics was also an important trait with 26.98% relative importance.
The trait that contributed less to diversity, were seed size per width and seed size per length. Tayade et al. (2015) obtained divergent results, where the characteristics that contributed least were hull content and weight of 100 seeds, with 0.00 and 0.05, respectively.

Conclusion
The safflower genotypes analyzed presented genetic divergence regarding the agronomic traits and the highest dissimilarity were PI248620 and PI343783, on the other hand, the less divergent genotypes were PI262450 and PI305198. The Tocher clustering and UPGMA hierarchical methods were partially concordant in ordering the similar accessions. The characteristics yield per plant and chapter number per plant are the ones that contributed the most for genetic dissimilarity in the safflower genotypes evaluated in the present research.