Genetic diversity and population structure of traditional sweet cassava accessions from Southern of Minas Gerais State, Brazil, using microsatellite markers

1 Programa de Pós Graduação em Genética e Melhoramento, Universidade Estadual de Maringá, Av. Colombo, 5.790, CEP 87020-900, Maringá, PR, Brasil. 2 Instituto de Investigação Agrária de Moçambique, Av. das FPLM 2698, C.P. 2698, Maputo – Moçambique. 3 Empresa de Pesquisa Agropecuária e Extensão Rural de Santa Catarina -Epagri /Estação Experimental de Caçador: Rua Abílio Franco, no 1500, C.P. 591, CEP 89500-000, Caçador, Santa Catarina, Brazil.

According to data recently published by FAO (2014), cassava is pointed out as crop that serves as nutrient source for more than 800 million people from the tropical and sub-tropical regions.African countries are currently leading the sweet cassava production as they possess most of the cultivated areas, such as Nigeria, the biggest world producer, followed by the Asian countries of Thailand and Indonesia (FAO, 2014).Brazil is the fourth biggest producer of sweet cassava in the world with a harvested area of 2,342.619ha and yield production of 23,087.828tons (IBGE, 2014).In a regional context, the yield production of cassava tuberous root in the year of 2013 in Minas Gerais State was estimated in 820,000 tons in a harvested area of 59,000 ha, occupying this way, the second place as the largest producer of cassava in the Southeastern Brazilian region, surpassed only by the state of São Paulo.
Cassava in Brazil is mainly grown on smallholdings by low-income farmers for their own consumption, a condition that ensures great genetic diversity to this crop (Zuin et al., 2009).This kind of subsistence farming has been extinguished in the past years due to the migration from rural to urban areas, and also to the expansion of large-scale farmers that now dominate areas previously occupied by smallholders, consequently leading to a drastic reduction of genetic diversity (Cleveland et al., 1994).Therefore, the characterization of the available germplasm, elucidation of genetic identity and population structure of sweet cassava accesses is very important (Costa et al., 2013;Ortiz et al., 2016).Genetic diversity can be assessed by using biochemical and morphological markers (Elias at al., 2001;Ortiz et al., 2016), but molecular markers are highly recommended because they provide more detailed information about polymorphism, independently of physiological state of the plant and environmental conditions (Agarwal et al., 2008;Raji et al., 2009).Molecular marker techniques, such as microsatellites (Raji et al., 2009;Monteiro-Rojas et al., 2011;Asare et al., 2011;Turyagyenda et al., 2012;Costa et al., 2013;Mezette et al., 2013;Kawuki et al., 2013;Ndung'u et al., 2014;Fu et al., 2014;Ortiz et al., 2016) have showed satisfactory results for evaluating the genetic diversity and population structure in sweet cassava.Furthermore, microsatellite markers, also known as simple sequence repeats (SSR), are informative due to their multi-allelic nature, codominant inheritance and wide distribution through the genome (Varshney et al., 2005;Raji et al., 2009).These tools require a low cost and are highly reproducible, which make them attractive and applicable for breeding programs (Turyagyenda et al., 2012).
The objective of the present work was to study the genetic diversity and population structure of sweet cassava accessions from South of Minas Gerais State, Southeastern of Brazil, using microsatellite markers.
At the instance of collection, the traditional sweet cassava accessions were catalogued and phenotypic variability among plants were registered.Plants with different characteristics located at the same site were collected separately.Approximately five to eight branches of 0.8 m from each mature adult plant were collected and individually identified.At the end of the collection a total of 50 accessions were obtained, which were maintained ex situ at the Sweet Cassava Germplasm Bank located at the Fazenda Experimental de Iguatemi (Iguatemi Experimental Farm), 23°S 20' 48'', 52°W 04' 17'') of the Universidade Estadual de Maringá (Maringá State University -UEM), PR, Brazil.The list of the evaluated accessions is shown in Table 1.
The planting was performed in February 2013.The stems were cut in small fragments of 0.15 to 0.20 m and planted at 0.10 m depth.For comparison purposes, the commercial sweet cassava cultivar IAC 576-70 was added as a control.The cultivar IAC 576/70 has some interesting features, such as resistance to the causal agent of bacteriosis (Xanthomonas axonopodis pv.manihotis), high yield in relation to other sweet cassava varieties, besides great biochemical and commercial characteristics (external root structure, cooking time, low content of HCN acid, yellowish root pulp), which fulfills the requirements of consumers in Brazilian Southeastern region (Vilella et al., 1985).At the end the total of access evaluated in this study were 51.

DNA extraction
The molecular analysis were conducted at the Laboratório de Melhoramento do Feijão Comum e de Biologia Molecular of the Núcleo de Pesquisa Aplicada a Agricultura (Nupagri) at the UEM (23ºS 26' 8'', 51ºW 53' 42").Genomic DNA extraction from freshly young leaves of each accession was performed following the protocol described by Dellaporta et al. (1983).The obtained genomic DNA was quantified using FluorometerQubit ® (Invitrogen, California, USA).DNA samples were diluted to obtain a final concentration of 50 ng µL -1 .
A set of 20 microsatellite markers identified and characterized in previous studies (Chavarriaga-Aguirre et al., 1998;Mba et al., 2001, Kunkeaw et al., 2010;Sraphet et al., 2011) were used to genotype the sweet cassava accessions (Table 2).PCR reactions were performed using 50 ng µL -1 of DNA; 0.25 mM of each deoxyribonucleotide (dATP, dCTP, dGTP and dTTP); 1.5 mM of MgCl2; 10 mM of 10x PCR Buffer, 0.08 μM of each primer (forward and reverse), 1 U of Taq DNA polymerase enzyme and add ultrapure water to achieve a total volume of 25 µL.
Thermal parameters for SSRY primers started of 94°C by 5 min followed by 30 cycles of denaturation at 95°C for 1 min, annealing temperature established for each primer for 2 min, extension at 72°C for 2 min and final extension for 5 min (Mba et al., 2001).
Further, amplification reaction for GA primers run as follows: initial denaturation at 94°C for 2 min; 30 cycles of denaturation at 94°C for 1 min, annealing temperature varying between 45 and 59°C, extension at 72°C for 2 min, and final extension at 72°C for 5 min (Chavarriaga-Aguirre et al., 1998).The amplification products were analyzed on 10% polyacrylamide gel stained with 0.02% SYBR Safe (Invitrogen, Oregon, EUA).The DNA bands were visualized under ultraviolet light and images were digitalized using the L-Pix EX photo documentation system (Loccus Biotech, Cotia, Brazil).

Data analysis
The population structure analysis of 51 sweet cassava accessions was performed using the software Structure 2.3.4 (Pritchard et al., 2000) based on a Bayesian algorithm to categorize the accessions into K groups.The analysis was conducted by assigning 10,000 interactions as burn-in and performing 100,000 interactions using the Markov Chain Monte Carlo (MCMC) method and 14 grouping simulations with the K-factor (number of groups in the population) ranging from 2 to 20.To determine the optimal K groups, the "ΔK" parameter was estimated using the Structure Harvester program (Earl and von Holdt, 2012), as proposed by Pritchard et al. (2000) and Evanno et al. (2005).
The software GenAlEx 6.5 (Peakall and Smouse, 2012) was used to estimate the allele frequencies per locus and to perform a principal coordinate (PCoA) analysis and analysis of molecular variance (AMOVA) (Excoffier et al., 1992).The statistical significance of the variances was tested using 999 random permutations at 1%.The genetic diversity of genes and alleles, the allele frequencies and the polymorphism information content (PIC) were calculated for each microsatellite locus using the Power Marker 3.25 software (Liu and Muse, 2005).A matrix based on the CS Chord distances (Cavalli-Sforza and Edwards, 1967) was used to develop a Neighbor-Joining Tree in the software Mega version 6 (Tamura et al., 2011).

Genetic diversity indices
In the present study, 68 different alleles were identified based on the analysis of 20 microsatellite markers, with an average of 3.4 alleles per marker (Table 3).A study developed by Chavarriaga-Aguirre et al. (1998) revealed the presence of 128 alleles in 522 sweet cassava accessions from Colombia (CIAT), showing a range from one to 15 alleles per marker.Later, Fregene et al. (2003)  All evaluated molecular markers were polymorphic and the average of the most frequent allele was 0.6056 (Table 3).The marker GA 131 showed the presence of five alleles.On the other hand, the molecular markers GA 136, SSRY 100 and SSRY 51 revealed the presence of two alleles (Table 3).The obtained results were consistent, corroborating to most studies regarding the use of SSR markers to evaluate the genetic diversity in sweet cassava and confirmed that approximately 100% of the tested markers are typically polymorphic (Siqueira et al., 2009;Turyagyenda et al., 2012).Although Chavarriaga-Aguirre et al. (1998) reported that the marker GA 57 was not polymorphic when tested in Colombian sweet cassava accessions, in the present study the GA 57 marker revealed the presence of three alleles.
In Costa Rica, a genetic diversity study including cassava accessions collected in indigenous communities has shown the presence of rare alleles for GA 140 marker (Rocha et al., 2008).A study carried out with 41 sweet cassava accessions from different Brazilian regions has further demonstrated the existence of rare alleles for GA 126 marker (Siqueira et al., 2009).In addition, genetic diversity studies performed with cassava accessions from Uganda, Africa (Turyagyenda et al., 2012) showed the presence of rare alleles for SSRY 19, SSRY 135 and SSRY 21 markers.A recent study that investigated the genetic diversity of traditional sweet cassava accesses from South region of Brazil revealed that GA 126 and GA 140 markers exhibited the occurrence of rare alleles (Costa et al., 2013).
On the other hand, the markers GA 136, SSRY 100, SSRY 101, SSRY 27, SSRY 35 and GA 134 showed the presence of high allele frequency which were respectively: 0.9489, 0.8653, 0.8600, 0.7708, 0.7446 and 0.7282 (Table 3).These high values could be directly related to the tendency of these alleles to fix in the population through increasing the homozygosis, since the  heterozygosity values for these alleles were low (Table 4).According to Nei (1987) these values are important to emphasize the genetic diversity among accessions and essential for evolutionary studies, because genetic change in the allele composition of the population could be evaluated by its allele frequencies.
The lowest allele frequency in this study was 0.0096, which was observed by the marker GA 126.However, this marker showed high heterozygosis value of 0.9423, which emphasizes the existence of allelic variation and, consequently, genetic diversity among the sweet cassava accessions (Costa et al., 2013).
The average of heterozygosity observed for the 20 molecular markers used in this study was estimated in 0.6487, which was considered high due to the fact that cassava plants are able to perform natural crossed fertilization, open pollination and protogynous flowering (Fregene et al., 2003).The mean heterozygosis value reported in this work is similar to previous population structure studies of cassava, which involved the use of microsatellite markers (Monteiro-Rojas et al., 2011;Asare et al., 2011;Turyagyenda et al., 2012;Kawuki et al.,  2000), as obtained using the program Structure Harvester (Earl and von Holdt, 2012).
The microsatellites that detected higher genetic diversity among the cassava accessions were GA 127, SSRY 21 and SSRY 28, with values of 0.7270, 0.6809 and 0.6185, respectively (Table 3).On the other hand, the markers GA 136, SSRY 100 and SSRY 101 exhibited the lowest values for genetic diversity, which were 0.0968, 0.2330 and 0.2434, respectively (Table 3).
Regarding the polymorphic information content (PIC), our results showed a mean value of 0.4080 (Table 3), which is moderately informative according to the methodology proposed by Botstein et al. (1980).The mean PIC value found was low compared to others studies related to cassava reported in the literature (Xia et al., 2005;Asare et al., 2011;Turyagyenda et al., 2012).On the other hand, studies developed by Costa et al. (2013) and Ortiz et al. (2016) with traditional sweet cassava cultivars from Paraná State, Brazil, showed PIC values such as 0.4040 and 0.4598.This fact can be explained because smaller and more homogeneous populations tend to present lower values of informative content per locus.

Population structure
The population structure analysis of the 51 traditional sweet cassava accessions according to Pritchard et al. (2000) revealed the formation of four distinct groups, according to Delta K value (Figure 2).The presence of admixture among the four sub-populations was observed.
The Figure 3 shows the distribution of the 51 sweet cassava accessions into the four sub-populations.The Group 1 allocated seven accessions, while the Group 2 was the most expressive containing 24 accessions, followed by the Group 3 and Group 4, which each one allocated 10 accessions.
The distribution dynamics was heterogeneous, as it was observed that the accessions from the same county or locality, such as BGM 643, BGM 642 and BGM 650 (from Alfenas, MG), were allocated in distinct groups through the probabilistic methodology (Pritchard et al., 2000).Similar results were obtained by Costa et al. (2013) and Ortiz et al. (2016) when these authors analyzed traditional sweet cassava accessions from Paraná State.This fact emphasizes the influence of exchanging cassava germplasm among cassava producers, upon population structure: accesses that were grown in different and distant regions were clustered in the same group (Lokko et al., 2006;Kizito et al., 2007, Ortiz et al., 2016).
These combinations are interesting because in addition to the divergence of the plants, most of these accessions are highly resistant to bacteriosis, caused by X. campestris pv manihotis, which is one of the most important cassava´s disease in Brazil, mainly in the Southern of Minas Gerais State where the accessions were collected.On the other hand, accessions BGM 669×BGM 668, BGM 670×BGM 671, BGM 652×BGM 653 were the most similar combination.This genetic similarity could be a reflection of the intense flow of material among the smallholders, generating duplicates (Kizito et al., 2007;Ortiz et al, 2016).
There was a discrepancy in the results when compared to the distribution of the accession inside of the formed groups by both methodologies proposed.In Group 1, the accession BGM 683 was the only one that clustered to the same group according to both methodologies.
According to the probabilistic methodology (Pritchard et al., 2000), the cultivar IAC 576/70 (BGM 686) was allocated in the group K = 2 (Figure 3), among two accessions from Poços de Caldas and Botelhos, MG.This results revealed genetic similarity between the cultivar IAC 576/70 and the accessions of the Group 2 (K=2) from Botelhos and Poços de Caldas.It is known that the IAC 576-70 comes from the cross between the clone IAC 14-18 and the traditional farming SRT 797-Ouro do Vale (Golden Valley), which was commonly found in small crops of cassava from Southeastern region of Brazil (Vilella et al., 1985).This fact probably contributed to the similarity of IAC 576-70 with Botelhos and Poços de Caldas accessions, which are very close to the border of Minas Gerais and São Paulo states.
Data regarding the analysis of molecular variance (AMOVA) is shown in Table 4.Our results revealed that a variance intra-groups was very expressive with a value of 93%, confirming the presence of a significant genetic diversity among the evaluated traditional cassava accessions.
The expressive value for genetic variance in these accessions reinforces what others mentioned about the cassava genetic nature being heterotic (Fregene et al., 2003).In Africa, using 98 cassava accessions from Uganda, Abaca et al. (2013) found an intra-group genetic variation (98%).Whereas in Brazil, Costa et al. (2013) analyzed the genetic diversity among 66 sweet traditional accessions from the Paraná region (Maringá, PR) and reported a value for the variation intra-groups of 77%.Another study of genetic diversity involving traditional sweet cassava accessions collected in the North of Paraná State showed 83% variation intra-groups (Ortiz et al., 2016).All these values of the variation intra-groups were similar to our results.
The low variation within the groups indicates that the accessions in each group are highly similar to each other but differ from the accessions included in the other groups.The differentiation among the four corresponding groups can also be compared by the Fst index, which estimates the genetic differentiation among groups.
According to Wright (1978), the PhiPT (analogous Wright Fst) coefficient is an indicative of genetic variability level among groups.The present study revealed a FST value of 0.073 that according to Wright's classification is considered of low genetic diversity (Table 4).Similar results of Fst index (0.07) were found by Ferreira et al. (2015), when the authors analyzed the genetic diversity of landraces of sweet cassava from the Mato Grosso do Sul State.Lokko et al. (2006) conducted studies of the genetic diversity in cassava accessions resistant to CMV (cassava mosaic virus).The authors found a values of Fst index that were considered low, while Ortiz et al. (2016) accessions of cassava from Maringá, Cianorte and Toledo, Parana state, verified a moderate diversity between the two groups of cassava accessions originally from Maringá, Paraná State (Fst index = 0.107).
The literature also shows other studies that exhibited low values for the Fst index, indicating low genetic variation among accessions formed in the population.For instance, Turyagyenda et al. (2012) obtained a value of 0.025 for genetic diversity between the two groups formed by the cassava accessions from Uganda.In addition, Mezette et al. (2013) reported that traditional accessions of cassava from different Brazilian regions revealed a low value of Fst equivalent to 0.041.It is plausible that a very low Fst index for cassava accession indicates a reduced overall genetic differentiation due to the random mating between groups or at least among their ancestors (Turyagyenda et al., 2012).
The Principal Coordinate Analysis (PCoA) contributed to the interpretation of the results for genetic diversity and population structure (Figure 5).The First Principal Coordinate (PCoA1) explains 26.33% of the variation among the accessions and is responsible for separating the accessions in four groups.The Second Principal Coordinate (PCoA2) explains 21.88% of the total variation.Together, these results are able to explain 48.21% of the genetic variation.

Principal coordinates analysis
Even though there is a certain divergence considering de grouping of the accessions among de different methods used, the individuals considered most divergent such as BGM 690 and BGM 655 were located in different regions of Figure 4.The same can be observed with the most similar accessions that are grouped together.It shows that the methods used had a certain convergence of results.

Conclusions
Our results showed that there was a moderate genetic diversity among the traditional sweet cassava accessions collected in the Southern of Minas Gerais State, Brazil, based on the SSR molecular markers.The PhiPT index was 0.073.The collected accessions were distributed in four distinguished sub-populations based on Bayesian analysis and multivariate analysis.The results revealed that the most dissimilar genotypes were the accessions BGM 690, BGM 655 and BGM 660, and the combination of such genotypes could provide a heterotic effect and transgressive segregation, ideally suited for cassava breeding programs.

Figure 1 .
Figure 1.Geographic localization of the counties from Southern of Minas Gerais State, Brazil, where the traditional sweet cassava accessions were collected.Source: IBGE(2016).

Figure 2 .
Figure 2. Inference of the number of K groups for the 51 sweet cassava accessions from Southern region of Minas Gerais State, Brazil, according to Pritchard et al. (2000), as obtained using the program Structure Harvester (Earl and von Holdt, 2012).

Figure 3 .
Figure 3. Population structure of 51 accessions of sweet cassava collected in Southern of Minas Gerais State, Brazil, assuming K = 4, according to Structure 2.3.4(Pritchard et al., 2000).

Figure 4 .
Figure 4. Distribution of 51 sweet cassava accessions from Southern region of Minas Gerais State, Brazil, based on 20 SSR loci obtained by the Neighbor Joining Tree from the genetic distance matrix of CS Chord(Cavalli-Sforza and Edwards, 1967), assuming K = 4 groups(Pritchard et al., 2000).

Table 1 .
Identification and origin of the sweet cassava accessions collected in the Southern region of Minas Gerais State, Brazil.

Table 2 .
Features of the microsatellite markers used to genotype the 51 sweet cassava accessions collected in the Southern region of Minas Gerais State, Brazil.

Table 3 .
Genetic diversity among 51 sweet cassava accessions from Southern region of Minas Gerais State, Brazil, using 20 SSR markers.

Table 4 .
Molecular analysis (AMOVA) of the sweet cassava accessions from Southern of Minas Gerais State, Brazil based on SSR markers.