GENETIC DIVERSITY IN AMBURANA (Amburana cearensis) ACCESSIONS: HIERARCHICAL AND OPTIMIZATION METHODS

The evaluation of accessions in a germplasm bank is essential for determining the potential parents in conservation programs, especially for native trees. This study aimed to determine the genetic diversity among 68 Amburana cearensis genotypes from diff erent locations in the state of Pernambuco, Brazil. Their genetic patterns were evaluated by Inter Simple Sequence Repeat (ISSR) molecular markers and genetic divergence was evaluated through multivariate analyses using diff erent clustering methods. The optimization method used (Tocher) was in agreement with all the hierarchical models used, in which clustering of the genotypes occurred similarly, specifi cally for the accession BB116, which is an important genetic material to be preserved and studied. Among the various hierarchical methods applied, the Average Linkage method exhibited higher discrimination power, allowing identifi cation of a larger number of divergent groups, thus implying wide genetic diversity among A. cearensis accessions.


INTRODUCTION
Amburana cearensis (Allem.) A. C. Smith (Fabaceae) or amburana is a tree native to the Caatinga (Brazilian xeric shrubland and thorn forest) with multiple uses, such as lumber; medicinal uses, attributed to coumarin, the active ingredient present in bark and roots; ornamental use; reforestation in agroforestry systems (Pimentel and Guerra, 2015); and forage. It is also important for bee raising as it provides nectar in the dry season of the year . All these features establish A. cearensis as a species of great socioeconomic and environmental importance in the locations in which it develops. In Brazil, it is found at altitudes that range from 20 -800 m in regions where average annual rainfall values can range from 500 to 1700 mm and temperatures from 19 to 29 °C (Carvalho, 1994). It can reach up to 20 m in height and 50 cm in trunk diameter and is characterized by pink fl owers and dark, fl at seed pods. Its seeds are dark and winged (Lorenzi and Matos, 2008). It has sexual propagation and a considerable percentage of germination (Angelim et al., 2007).
Evaluation of genetic diversity is essential for breeding and conservation of species. Reduction in genetic diversity makes a species more susceptible to environmental changes, reducing its chances of survival over time (Spigler et al., 2017). Knowledge of genetic diversity of A. cearensis becomes crucial in identifi cation of parentage for establishing a breeding and conservation programs, providing conditions for development of cultivars as well as discovery of more promising genotypes in medicinal and adaptive aspects. However, collecting this information is still in the initial stage, leading to the need for studies regarding A. cearensis. Molecular markers are useful tools to determine the extent of genetic diversity of populations of plants (Borém and Caixeta, 2016). Molecular techniques such as Inter Simple Sequence Repeats (ISSR) that allow identifi cation of wide intra-and inter-specifi c diversity are used in studies due to precise evaluation (Lorenzoni et al., 2014) and their other advantages, such as quick performance, cost-benefi t ratio, and small amount of DNA template. In addition, they do not require previous knowledge of target sequences (Ng and Tan, 2015;Grover and Sharma, 2016). Consequently, ISSR is an eff ective tool for a variety of purposes, including identifi cation of germplasm and evaluation of genetic diversity (Coutinho et al., 2014;Kumar and Agrawal, 2017;Cid-Contreras et al., 2019;Al Salameen et al., 2020;Xiang et al., 2020).
The analyses performed from molecular markers include clustering methods. The purpose of these methods is to separate an original group of observations into various subgroups in order to obtain homogeneity within the subgroup and heterogeneity among the subgroups. Genotypes are clustered by a process that is repeated on various levels, and a dendrogram is established, without concern for the optimal number of groups. This is suitable for identifying divergent genotypes with the greatest probability of success in crosses (Cargnelutti Filho et al., 2008). In optimization methods, groups are established by refi ning a determined clustering criterion, diff ering from hierarchical methods by the fact of the groups formed being mutually exclusive. In addition, diverse methods based on diff erent measures of dissimilarity can lead to distinct clustering patterns (Cruz et al., 2014).
Studies that deal with genetic diversity of A. cearensis involving diff erent clustering methods are still quite limited. Therefore, the aims of this study were to characterize the genetic divergence among 68 A. cearensis accessions by means of molecular traits; determine if there is consistency between the Tocher clustering method and diff erent hierarchical methods (Gower, Average Linkage, UPGMA, Ward, Complete Linkage, and WPGMA); and indicate promising hybrid combinations for obtaining potentially important populations for A. cearensis breeding and conservation programs.

MATERIAL AND METHODS
The study was conducted from Each accession was sown in seedling plugs containing a soil and sand mixture (1:1). Seven days after germination, the seedlings were transferred to 1000 mL capacity plastic bags containing the same substrate used in the previous step. Six months after transplanting, when the plants reached around 50 cm in height, molecular analyses began. Genomic DNA was obtained from young leafl ets collected from plants kept in a greenhouse according to the method proposed by Nunes et al. (2011). Five ISSR primers recommended in the literature as polymorphic were used with the respective sequences (5'-3') and annealing temperatures (Ta): S-31-AGAGAGAGAGAGAGAGVC (49.6 ºC); UBC812-GAGAGAGAGAGAGAGAAA (51.6 ºC); UBC827-ACACACACACACACACG (50 ºC); UBC890-VHV GTGTGTGTGTGTGTT (52 ºC); and UBC891-HVH TGTGTGTGTGTGTG (52 ºC; 51.6 ºC).
The ISSR amplifi cation reactions were performed in a fi nal volume of 25 µl, containing 50 ng of DNA, 7.5 µl of 5X reaction buff er, 1.5 mM of MgCl 2 , 200 µM of each dNTP, 0.8 µM of the primer (Sigma, USA), and 2.4 U of Taq DNA polymerase (Go Taq Flexi, Promega, USA). The reactions were conducted in a gradient thermal cycler (Multigene Gradient, Labnet International, USA) programmed for an initial denaturation step of 2 min at 95 °C, followed by 40 cycles of denaturation at 95 °C for 45 sec. Annealing temperature was defi ning in accordance with each primer. Both annealing processes occurred for 1 min and extension of the primers was at 72 °C for 2 min. A fi nal extension step was carried out at 72 °C for 5 min.
The bands generated by amplifi cation of the genomic DNA from each accession were used as data for interpretation in this study. The bands amplifi ed by the same primer that occupied the same position on the gel were considered as belonging to the same gene loci. In accordance with the same principle, bands generated by the same primer but that occupy diff erent positions were considered as belonging to diff erent gene loci. Thus, the ISSR markers were genotyped regarding the presence (1) and the absence (0) of the bands, generating a binary matrix, used for division of the variance among their components within and among populations (hierarchical levels).
Analysis of variance was performed on the data by the F test (P= 0.05) and the means were compared by the Scott-Knott test (P= 0.05). After that, multivariate analyses were performed for the purpose of determining genetic dissimilarity among the genotypes, obtaining the dissimilarity matrix by hierarchical methods. Genetic divergence was represented by dendrograms obtained by six hierarchical methods: Gower, Average Linkage, UPGMA, Ward, Complete Linkage, and WPGMA, and by the Tocher optimization method. The clustering by hierarchical methods was validated through the cophenetic correlation coeffi cient (CCC), calculated by the test of Mantel (1967). All the data obtained were analyzed using Genes v. 2015.5.0 software (Cruz, 2013).

RESULTS
The means test of the molecular pattern of the A. cearensis populations showed wide genetic diversity within the sampled populations themselves that exhibited distinct individuals in regard to DNA content (Table 1).
The cophenetic correlation coeffi cients for the diff erent hierarchical methodologies (Gower: 0.84, Average Linkage: 0.48, UPGMA: 0.89, Ward: 0.43, Complete Linkage: 0.74, and WPGMA: 0.80) exhibited a suitable relationship between the distance matrix and the dendrogram generated, which shows fi delity in presentation of the dataset (Figure 2). Thus, the genotypes were clustered in three to six distinct groups, when a cut was made considering 10% of dissimilarity. In all the hierarchical methods, Group I was represented only by the accession BB116, and for the other groups, genotype distribution was variable (Table 2). The Average Linkage method provided the best representation of the dendrogram, since it allowed better formation of subgroups, i.e., this method was able to distinguish the accessions in a more detailed way, classifying them with higher diversity, which is a relevant result for breeding purposes. In contrast, the UPGMA method generated the least satisfactory representation in cluster formation, detecting less genetic diversity among the accessions and, thus, dendrograms that are less useful depending on the desired purpose.
Cluster analysis by the Tocher method among the 68 accessions led to the formation of two groups, such that within each group, the matrices were considered to be genetically similar, and between groups, dissimilar (Table 3). Approximately 98.5% of the genotypes were concentrated in Group I, and the genotype BB116 alone composed Group II.

DISCUSSION
The genetic diversity within the A. cearensis populations can provide an estimate of apparent gene fl ow. Tree species generally have a mixed mating system, as well as long-distance pollen and seed dispersal, which favors the occurrence of gene fl ow among populations, thus increasing the genetic diversity within populations and decreasing the genetic diff erentiation among them (Zanettini and Cavalli, 2003). Species with pollination agents that reach great distances (such as wind, birds, or bats) and/or dispersers that distribute seeds across extensive areas (such as wind or large animals) have highest gene fl ow and consequently high genetic diversity within populations (Mori, 2003). Amburana cearensis is pollinated by insects (bees, fl ies, and moths) and seeds are wind dispersed (Araujo and Dantas, 2018). This information contributes to an understanding of this great genetic diversity among accessions.
The determination of groups of parents for hybridization can be determined based on genetic divergence among individuals (Martins et al., 2002). The most divergent individuals are the most suitable for performing crosses since they may produce a greater heterotic eff ect and genetic diversity in the segregating generations. This information is highly worthwhile for the species in question, which is still in a wild state in terms of domestication, and this type  of directing serves as an essential tool for eff ective search by this process.
In analyzing the aspect of plant selection for seed collection, genetic diversity is fundamental for ensuring genetic gains by selection in plant breeding (Voss-Fels et al., 2019). In this case, the analyzed Table 2 -Division of Amburana cearensis genotypes in diff erent genetic groups based on six hierarchical clustering methods. Tabela 2 -Divisão de genótipos de Amburana cearensis em diferentes grupos genéticos baseada em seis métodos hierárquicos de agrupamento. Genotypes  method  I  BB116  II  BB115 populations indicated potential for this purpose, since intrapopulational genetic diversity was satisfactory, conferred by diversity of groups shown in the six clustering methods applied. Similar results with application of similar methodologies were also reported in populations of Spondias lutea (Gois et al., 2014a) and Dimorphandra mollis (Oliveira et al., 2008).  Group  Genotypes  I  RO93, RO98, BJ76, J52, J41, J20, J57, J55, BB124, J53, BJ88, RO104, J51, J43, BJ81, J54, RO90, J27, RO94 According to McKay et al. (2005), the genetic diversity within populations is important because the collection of material from diff erent forest environments that have edaphic and climatic characteristics similar to that to be restored have a direct eff ect on eff orts for restoration and conservation of native species. Exotic genotypes can reduce the success of these types of eff orts through poor adaptation to local conditions or through reduction on the gene fl ow of adjacent native populations.

Hierarchical Group
The genotype BB116 representing an important accession that merits more attention since it has the genetic identity with higher divergence from the other individual's analysis by the Tocher method. This response may indicate that this accession has a diff erent origin from the others. A similar result was found in Phaseolus vulgaris (Cargnelutti Filho et al., 2008), in which application of this optimization method separated the groups in a manner quite similar to that obtained in this study. Likewise, in a study with individuals of S. lutea, 80% of the accessions were clustered in one group alone by Jaccard genetic similarity (Gois et al., 2014a). Therefore, as the accession BB116 has these characteristics, it has considerable potential for use in crosses for generating hybrids. At the other extreme, information regarding the most similar pairs is useful in programs involving backcrosses, in which the use of similar parents (basically diff erentiated by the allele to be transferred) allows the recurrent parent to be recovered (Araujo et al., 2002). This information is highly relevant for conservation programs since it assists actions that allow domestication of A. cearensis with wide genetic diversity, together with selection of good adaptive and agronomic traits.
Application of the Gower, Average Linkage, UPGMA, Ward, Complete Linkage, and WPGMA hierarchical methods and of the Tocher optimization method indicated the existence of genetic variability among the genotypes. The number of groups formed by the Average Linkage method was higher (six groups) than that obtained by the other hierarchical methods, showing greater discriminating power. This allows identifi cation of more groups containing similar accessions.
The results generated in this study confi rm the importance of molecular markers for the proposed objective, and this study is in agreement with other studies reported in the literature (Jiang et al., 2019;Al Salameen et al., 2020;Xiang et al., 2020;Zacarías-Correa et al., 2020). The results also reinforce the importance of application of this technique in studies of other native tree species that still lack greater investigation in the scientifi c community and are exposed to extinction.

CONCLUSIONS
There is genetic diversity among the A. cearensis accessions, confi rmed by detection of diff erent groups formed by the clustering methods, suggesting that studies directed to crosses be performed among accessions of these diff erent groups. The Average Linkage hierarchical method exhibited greater discrimination power, allowing identifi cation of a larger number of groups containing similar accessions. The Tocher optimization method proved to be eff ective in representing the genetic distance of the accessions in this study, since it was in agreement with all the hierarchical methods, revealing that the accession BB116 is the most divergent of the accessions and is therefore, indicate for performing crosses and for use in conservation programs of native tree species.