Genetic characterization of cassava (Manihot esculenta) landraces in Brazil assessed with simple sequence repeats

Based on nine microsatellite loci, the aim of this study was to appraise the genetic diversity of 42 cassava (Manihot esculenta) landraces from selected regions in Brazil, and examine how this variety is distributed according to origin in several municipalities in the states of Minas Gerais, São Paulo, Mato Grosso do Sul, Amazonas and Mato Grosso. High diversity values were found among the five above-mentioned regions, with 3.3 alleles per locus on an average, a high percentage of polymorphic loci varying from 88.8% to 100%, an average of 0.265 for observed heterozygosity and 0.570 for gene diversity. Most genetic diversity was concentrated within the regions themselves (HS = 0.52). Cluster analysis and principal component based scatter plotting showed greater similarity among landraces from São Paulo, Mato Grosso do Sul and Amazonas, whereas those from Minas Gerais were clustered into a sub-group within this group. The plants from Mato Grosso, mostly collected in the municipality of General Carneiro, provided the highest differentiation. The migration of human populations is one among the possible reasons for this closer resemblance or greater disparity among plants from the various regions.


Introduction
Traditional or slash-and-burn farming is a system with characteristics related to the pre-colonial period, and which is preserved by both indigenous and other populations that employ techniques transmitted culturally by their ancestors (Faraldo et al., 2000). The basic evolutionary unit of traditional farming, the "swidden field", is where both in situ conservation of landraces from many species of economic importance and genetic amplification of diversity occur, with subsequent benefits to the farmer (Martins, 1994(Martins, , 2001Peroni and Martins, 2000;Sambatti et al., 2001). The terms landrace, ethno-variety and folk or local variety define plant populations which are ecologically or geographically distinct, and are differentiated in their internal genetic composition, as a result of local selection by traditional farmers (Brown, 1978). Traditional agro-systems are of particular interest as they usually represent high crop diversity. It is common to find numerous varieties in the same field (Elias et al., 2000). Some of these species have received attention by researchers through the focus on genetic characterization, for example sweet potato (Veasey et al., 2007(Veasey et al., , 2008, yam (Malapa et al., 2005), taro (Jianchu et al., 2001), maize (Louette et al., 1997), bananas (Creste et al., 2003), and most important of all, cassava (Sambatti et al., 2000;Mühlen et al., 2000;Faraldo et al., 2000;Peroni, 2007).
Cassava (Manihot esculenta Crantz), known in Brazil as "mandioca", "macaxeira" or "aipim", is the main crop cultivated in traditional farming systems in Brazil, as well as other areas in tropical America (Martins, 1994). It is an important subsistence crop for many communities with flexible planting and harvest times . Represented by a high number of varieties, cassava stands out as a suitable model for analyzing the inter-relationship between societies, genetic resources and ecological conditions. Studies concerning cassava diversity are scarce when compared with the great ethnical and territorial diversity of the populations that grow M. esculenta. Socio-cultural contexts, as well as economic and ecological processes, exert an influence on the management of this crop with variable intensity. The high diversity observed in those traditional populations that cultivate cassava reflects a pre-and postcolonial history, consisting of migrations, inter-ethnic contacts and economic pressures (Emperaire et al., 2001). diversity in cassava landraces in Brazil and other countries, various studies have been undertaken with isozyme markers (Sambatti et al., 2000;Faraldo et al., 2000;Cabral et al., 2002;Resende et al., 2004), randomly amplified polymorphic DNA (RAPD) markers (Colombo et al., 1998;Carvalho and Schaal, 2001;Zacarias et al., 2004), and microsatellites or simple sequence repeats (SSR) (Mühlen et al., 2000;Carvalho and Schaal, 2001;Peroni, 2007), the latter being an appropriate marker for the detection of genetic polymorphisms, widely used to characterize genetic diversity in traditional crops (Mühlen et al., 2000;Fregene et al., 2003;Elias et al., 2004;Veasey et al., 2008).
In order to assess the genetic diversity of local varieties in farmers' homesteads and their distribution throughout different regions in Brazil, as a means of devising better conservation approaches and identifying progenitors with a wide genetic base for breeding, 42 landraces were evaluated with nine SSR markers. The plants were divided into five groups according to geographic origin: MG -Minas Gerais; SP -São Paulo; MS -Mato Grosso do Sul; AM -Amazonas; and MT -Mato Grosso. Certain groups, such as MS and MT, were either not represented or poorly so in the previous studies mentioned above. Thus, further information should be propitious concerning the genetic diversity of cassava in these regions.

Plant material
A total of 42 landraces from the cassava germplasm bank of the Genetics Department of ESALQ/USP, Piracicaba, SP, were assessed according to their geographical origin, with the aim of obtaining a representative sample from different parts of Brazil. These plants were collected from homesteads undertaking traditional farming methods, in several municipalities in five regions of Brazil, and were classified into five groups, according to their place of origin: MG -municipality of Frutal, Minas Gerais State; SPmunicipalities of Eldorado, Cananéia and Ilha Comprida, in the Vale do Ribeira, São Paulo State; MS -municipalities of Sonora, Pedro Gomes, Rio Verde de Mato Grosso, Costa Rica, Cassilândia, Paranaíba and Inocência, Mato Grosso do Sul State; AM -municipalities of Uarini, Maraã and Alvarães, in the Mamirauá and Amanã Sustainable Development Reserves, Amazonas State; and MT -municipality of General Carneiro, Mato Grosso State ( Figure 1; Table 1).

DNA extraction and quantification
DNA was extracted from recently expanded young leaves of each accession and then dehydrated for 72 h at 60°C by using a modified CTAB method (Elias et al., 2004). Fifty milligrams of ground powder were transferred to a 1.5 mL micro-tube containing 800 mL of CTAB extraction buffer [30 mM EDTA pH 8.0, 0.1 M Tris-HCL pH 8.0, 1.2 M NaCl, 3% CTAB, plus 1% 2-mercaptoethanol added just before use]. After incubation at 65°C for 1 h, 500 mL of chloroform-isoamylalcohol (24:1) were added and the mixture subsequently centrifuged at 8,000 rpm for 10 min. This step was repeated once again. The supernatant (400 mL) was then transferred to a fresh tube with 350 mL of -20°C isopropanol and stored at -4°C for 1 h, whereupon it was once more centrifuged at 8,000 rpm for 10 min. After drying, 200 mL of TE buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA) and 4 mL RNAse (10 mg/mL) were added to each tube. The DNA was quantified in 4% polyacrylamide gel electrophoresis by using the silver nitrate staining technique (Bassam et al., 1991), so as to visualize the DNA bands in the gels.

Microsatellite amplification
The amplification reaction was run on a total volume of 10.2 mL, consisting of 0.2 mL of Taq Polymerase (5 U/mL); 1.0 mL 10x Buffer ; 1.0 mL MgCl 2 (50 mM); 0.5 mL of each primer (F/R) (5pmoles/mL); 1.0 mL dNTP (2.5 mM of each deoxyribonucleotide); 3.0 mL Milli-Q H 2 O and 3.0 mL DNA (5 ng). Based on previous studies (Mühlen et al., 2000;Elias et al., 2004;Peroni et al., 2007), nine microsatellite primers were used (Chavarriaga-Aguirre et al., 1998) ( Table 2). PCR reactions were performed with a MWG-BIOTECH Primus 96 thermocycler, with the following sequences: 4 min at 95°C, 29 cycles of 1 min at 95°C; 2 min at the annealing temperature defined for each primer (Table 2); 2 min at 72°C; and a final extension stage of 1 min at 72°C. Separation of the amplified product was accomplished in 6% polyacrylamide gel electrophoresis at 60 V for 30 min and 120 V for 3 h and 30 min. The gels Siqueira et al. 105 were stained with silver nitrate (Bassam et al., 1991) and photo-documented.

Statistical analyses
Genetic diversity parameters, such as the number of alleles per locus, allelic frequency, percent of polymorphic loci, observed average heterozigosity and gene diversity (expected heterozygosity) obtained per locus and per group of accessions, were estimated using GDA software (Lewis and Zaykin, 2000). Allelic frequencies and Nei (1973) genetic diversity parameters were estimated with FSTAT software (Goudet, 2001).
Cluster analysis with the 42 landraces was performed with NTSYS software, by using binary data whereby alleles were transformed into the presence or absence of an SSR band, as well as with the Jaccard similarity coefficient and UPGMA (Unweighted Pair Group Method with an Arithmetic Mean) method. A principal component analysis with binary data was also carried out using the SAS (1999) program, BioStat 4.0 software (Ayres et al., 2005) providing a scatter plot.

Results
A total of 46 alleles were amplified with nine SSR loci analyzed in the 42 landraces, the number of alleles observed per locus varying from 3 to 6 alleles ( Table 2). The number of alleles per polymorphic locus in the five cassava groups varied from 2.1 to 3.8, with an average of 3.3 (Table 3). Specific alleles were detected in the MS, MT and SP groups. Some of these alleles were considered to be rare (0.050 frequency), such as alleles 3 and 5 for GA-5 in the MS group and allele 3 for GA-126 in that of MT.
The groups MS, AM and MT revealed 100% polymorphism, while those of MG and SP presented 88.8%. The observed heterozygosity, 0.265 on an average, varied from 0.233 to 0.288, while higher values, varying from 0.427 to 0.677, were estimated for gene diversity, with an average of 0.570.
According to Nei diversity indices (Nei, 1973), high total diversity (H T = 0.635) was observed in all the 42 cassava landraces, thus confirming the high variability which can be found in this cross-pollinated and vegetatively prop-106 Cassava landrace characterization  Table 4).
The dendrogram in Figure 2 shows the high genetic variability of landraces, this varying from 0.19 to 0.83 in the Jaccard similarity coefficient. Five groups were defined through cluster analysis. Landraces from all the five regions were classified in the first of these, the five accessions from Frutal (MG) being gathered into a small sub-group of this larger one. As to clustering in the other four groups, landraces only from General Carneiro (MT) were classified in the second group, four from Sonora and Pedro Gomes (MS), and two, one from Cananéia and the other from Ilha Comprida (SP) in the third, two from the Amazon region plus one from Inocência (MS) in the fourth and finally, only one single specimen from Cassilândia (MS) in the fifth.
Similar results were observed through principal component analysis when examining the distribution of the five groups in the four quadrants of a scatter plot compounded with the two principal components, and which explained Siqueira et al. 107

Discussion
In our study, high genetic diversity was detected in all the five regions in Brazil, with an average of 5.0 alleles per locus, which is in agreement to similar studies with cassava (Muhlen et al., 2000;Faraldo et al., 2000;Fregene et al., 2003;Mkumbira et al., 2003;Elias et al., 2004;Lokko et al., 2006). Peroni et al. (2007) evaluated 137 cassava specimens representing 58 landraces from Brazil, by using nine SSR loci, thereby reporting an average of 4.56 alleles per locus, this varying from 2 to 7 alleles. When analyzing 283 accessions from various countries with 67 SSR loci, Fregene et al. (2003) found an average of 5.02 alleles per locus for Brazilian landraces.
High gene diversity values of 0.570 on an average were also encountered in our study. By using SSR markers, gene diversity was found to be high in all the cluster groups of cassava analyzed by Lokko et al. (2006), with an average of 0.447. When assessing 283 accessions of cassava land-races from Africa and the Neotropics, Fregene et al. (2003) also came upon high gene diversity, 0.535 on an average with 67 SSR loci. On the other hand, on studying 137 cassava plants from 58 landraces in Brazil, Peroni (Peroni N, PhD Thesis, UNICAMP, 2004) obtained the even higher value of 0.637 for average expected heterozygosity or gene diversity. All these values are high when compared to the average gene diversity for outcrossing species of 0.205, estimated for all plant species, and 0.159 for dicots (Hamrick and Godt, 1997). These high values also substantiate both the cassava outcrossing breeding system, with multi-loci outcrossing rates estimated at 91.5% when using isozyme markers , as well as its highly heterozygous nature due to its vegetative mode of reproduction. In French Guiana, Pujol et al. (2005) noted a positive correlationship between plant size and heterozygosity, thus concluding that during weeding farmers tended to eliminate the less vigorous plantlets, which in itself could explain the higher levels of heterozygosity reported in the literature.
A larger portion of diversity in this study was found to be concentrated within the groups themselves (H S = 0.552) on the contrary to group diversity (G ST ' = 0.131). Faraldo et al. (2000), Mühlen et al. (2000) and Asante and Offei (2003) also observed that most morphological, isozymatic and molecular variability was concentrated within either the cassava 'swidden' fields cultivated by traditional farmers or geographic regions. Peroni et al. (2007), when using nine SSR loci to analyze 58 cassava landraces, arrived at similar results, thereby explaining that each farmer maintains an appreciable representation of total diversity in his homestead, this diversity not being different from that of other farmers in the same region, and is due to material exchange between relatives and neighbors. Lokko et al. (2006) also found that most gene diversity assessed with SSR markers was concentrated within cluster groups of cassava from Africa. The same pattern was observed with sweet potato landraces from the Vale do Ribeira, with greater genetic variability within swidden fields for both morphological (Veasey et al., 2007) and SSR markers (Veasey et al., 2008).
In addition to measuring genetic diversity, one of the objectives in our study was to verify how landraces originating from five different regions in Brazil were mutually related. Results showed a greater proximity of landraces from the states of São Paulo, Mato Grosso do Sul and Amazonas. Faraldo et al. (2000), on studying cassava landraces from three distinct groups (the Indigenous Park of Xingu, the Vale do Ribeira in São Paulo and the Amazon region), found greater likeness among landraces from São Paulo and the Amazon. Peroni et al. (2007) also identified greater genetic similarity among landraces from the Vale do Ribeira and those of the Rio Negro (Amazon). According to these authors, the samples from São Paulo, represented here by landraces from the Vale do Ribeira, may 108 Cassava landrace characterization represent a "historical sample" of Amazonian cassava diversity. The landraces from Minas Gerais were clustered into a sub-group in the dendrogram (Figure 2), within a larger group of landraces from São Paulo, Mato Grosso do Sul and Amazonas. The landraces collected in Frutal, Minas Gerais, an area dominated by soybeans, pineapples, sugar and pasture dedicated to milk and beef production, were all cultivated in home-gardens and not in 'swidden' fields. These are known as sweet-varieties and are used for homecooking. The origin of these landraces is apparently local, with plant material being exchanged among relatives, friends and neighbors and, according to the villagers, have been under cultivation in this area over a long period (Angelo and Amorozo, 2006).
The most differentiated landraces were those from General Carneiro, Mato Grosso. In Mato Grosso, traditional farming is undertaken by local populations, particularly by indians, "quilombolas" and "pantaneiros". After the 50's more recent settlers began coming from diverse regions of the country, mainly from the south, northeast, Goiás and Minas Gerais (Amorozo, 2000). General Carneiro is an important area for settlement, the high genetic variability of cassava landraces being a consequence of the introduction of genetic material from the settler's place of origin. This municipality presents certain peculiarities that may help to explain the differentiation of local landraces from those of other regions. More extensive modern farming with soybean and cotton crops, predominant in other parts of the state, is a reality for farmers in this area. Information on modern cassava varieties and released by plant breeding institutes in Brazil (Instituto Agronômico -IAC, Embrapa) can easily reach local farmers through the radio and satellite television. However, due to economical and transport limitations, access to this material is very rare. Some local farmers reported impediments in obtaining new varieties for homestead planting, even of those from the same area, due to the difficulty and high cost in transportation. Under local conditions, distances as short as 20 kilometers can constitute an isolation factor.
Both sweet and bitter varieties of cassava are grown by farmers in the States of São Paulo (Amorozo, 2000) and Mato Grosso do Sul, substantial diversity of sweet varieties being encountered in the latter (Zatarin M and Valle TL, personal communication). Sweet varieties have a low level of cyanogenic glucoside content in their roots, whereas the bitter type presents more than 100 ppm fresh weight of cyanide therein which must be detoxified before consumption (Valle et al., 2004). The Vale do Ribeira region, represented by Cananéia, Ilha Comprida and Eldorado, is an area with a high diversity of species and varieties, especially cassava, and it is possible that Tupi-Guarani populations were the main disseminators of both cultivation techniques as well as of the species and varieties to be found there (Peroni N, PhD Thesis, UNICAMP, 2004). It is possible that the greater similarity among landraces from São Paulo and Mato Grosso do Sul can be traced back to migrating populations coming from southern Amazonia, bringing varieties which were introduced into the Vale do Ribeira together with their traditional cultivation techniques (Schaden, 1974). Some cassava varieties were probably carried along with Tupi-Guarani indians migrating from São Paulo to communities in Mato Grosso do Sul (Ladeira MI, PhD Thesis, USP, 2001). Thus, these could be some of the factors contributing to an explanation of the proximity of landraces from São Paulo with those collected in the cerrado ecosystem of Mato Grosso do Sul and the tropical Amazon Forest.
In this study we intended to contribute to a better knowledge of cassava genetic diversity and distribution within and among different regions in Brazil, and promote in situ conservation by traditional farmers, also known as on farm in situ conservation, of an important genetic and autochthon resource.