Diversity within American cassava germ plasm based on RAPD markers

This work focuses on the genetic diversity of American cassava through RAPD molecular markers. The 126 genotypes studied were distributed on four geographical levels ranging from local to continental. Samples included ethnocultivars from the Santa Isabel community in the Brazilian Amazon, local cultivars collected in the State of Sao Paulo, native accessions from very diverse Brazilian regions, and representative accessions from the Centro Internacional de Agricultura Tropical (CIAT) core collection. Eighty-eight polymorphic bands were analyzed. Results revealed the weak genetic structure of the cassava analyzed. The pattern formed by the first two axes of the principal coordinates analysis (PCoA) revealed an overlapping of the Sao Paulo State genotype, the Brazilian group and the core collection accessions. The Santa Isabel ethnocultures formed a separate group. The weak genetic structure of cassava can be explained by the common practice of exchanging botanical material among small producers as well as by recombinations among genotypes. When the genotypes were analyzed using climatic data, the sample sites were found to be structured according to temperature and precipitation. RAPD markers proved very useful in the genetic diversity study, resulting in important implications for cassava germ plasm collections and genetic breeding.


INTRODUCTION
Cassava (Manihot esculenta Crantz, Euphorbiaceae) is one of the most important food crops in the tropics.It is a valuable source of carbohydrates, a staple in several developing nations in Africa, South America and Asia, and has the highest production potential of calories per hectare per day among tropical crops (De Bruijn and Fresco, 1989).Cassava is a shrubby species originated in the American continent approximately 3000 to 7000 years ago in several different regions: semi-arid and "cerrado" regions in Brazil (Rogers, 1972), savannas in the Orinoco River Basin between Colombia and Venezuela (Sauer, 1952), which originally belonged to adept cassava farming Aruak Indian tribes (Schmidt, 1951); hot, dry regions in Mexico and Guatemala (Rogers, 1972), and deserts along the Peruvian coast, which descended from the savannas created after glaciation (Ugent et al., 1986).
The large number of cultivars represented in the main germ plasm collections of several research centers indicates the high genetic diversity of this species.According to Martins (1994), cassava's great diversity can be explained by life history components and species dynamics as well as the horticultural practices of different Indian and traditional farmer ("caboclo" and "caiçara") communities.Cultivars are typically randomly arranged in traditional plots, which favors intercrossing.Moreover, the seeds produced in these plots disperse, fall to the ground and lie dormant for a long time.
Another possible source of new variation is mutations or interspecific crosses between wild and/or weedy Manihot species with those being cultivated.The wild and/ or weedy species may be found surrounding or inside the plot.Mutations or crosses could be fixed and propagated vegetatively (Martins, 1994).
According to Cohen et al. (1991), ex situ preserved cassavas represent about 16,000 accessions.Hershey (1994) estimated about 7,000 different cultivars.Furthermore, due to the large number of cultivars represented in collections, Boster (1985) proposed that some new cultivars were produced by human populations who traditionally consume this tubercle.The author cited new cultivars created by native populations of the Amazon Basin using visual selection criteria.Jeffrey (1968) stated that high polymorphism rates could be explained partly by vast, varied cultivation areas and conscious and unconscious breeding decisions, which accelerated this plant's evolution.
Molecular markers are a more stable and informative alternative to isoenzymes.According to Cohen et al. (1991), these markers can more efficiently be used to examine the genetic diversity of collections.In fact, many molecular markers are currently being used to study the genetic diversity of several different species.Marmey et al. (1994), Bonierbale et al. (1995) and Colombo (1998) have used molecular markers to study genetic diversity in cassava; however, only a limited number of genotypes were studied.

Abstract
This work focuses on the genetic diversity of American cassava through RAPD molecular markers.The 126 genotypes studied were distributed on four geographical levels ranging from local to continental.Samples included ethnocultivars from the Santa Isabel community in the Brazilian Amazon, local cultivars collected in the State of São Paulo, native accessions from very diverse Brazilian regions, and representative accessions from the Centro Internacional de Agricultura Tropical (CIAT) core collection.Eighty-eight polymorphic bands were analyzed.Results revealed the weak genetic structure of the cassava analyzed.The pattern formed by the first two axes of the principal coordinates analysis (PCoA) revealed an overlapping of the São Paulo State genotype, the Brazilian group and the core collection accessions.The Santa Isabel ethnocultures formed a separate group.The weak genetic structure of cassava can be explained by the common practice of exchanging botanical material among small producers as well as by recombinations among genotypes.When the genotypes were analyzed using climatic data, the sample sites were found to be structured according to temperature and precipitation.RAPD markers proved very useful in the genetic diversity study, resulting in important implications for cassava germ plasm collections and genetic breeding.
A survey of a large number of cultivars from different regions is needed to better understand the organization and extent of genetic diversity, structuring factors, degrees of relationship between genotypes and their characterization in cassava.Consequently, conservation strategies could be developed to better exploit this germ plasm in the future.This study used RAPD molecular markers to determine the genetic diversity of American cassavas from different ecological and geographical zones.

Material
One hundred and twenty-six assessions of M. esculenta were studied (Table I).Genotypes were grouped into the following four geographic levels, ranging from local to continental: a) 22 cassava ethnocultivars from a small community of "caboclos" in Santa Isabel, Amazon region (middle Negro River); b) 19 cultivars from the State of São Paulo; c) 52 accessions from different Brazilian geographical zones; d) 33 accessions from the CIAT collection selected to represent world diversity (based on a preliminary RAPD study).

DNA isolation and amplification
DNA was isolated from silica gel dried leaf material according to the following methodology: 0.5 g dry leaves were ground in liquid nitrogen and transferred to 20-ml plastic tubes.Ten milliliters of extraction buffer (0.1 M Tris HCl, pH 8.0, 1.25 M NaCl, 0.02 M EDTA, 2% MATAB (mixed alkyltrimethylammoniun bromide)) and 1% ßmercapto-ethanol was added just before use.After 90-min incubation at 65°C with slow stirring, an equal volume of chloroform/isoamylalcohol (24:1) was added twice and the resulting supernatant transferred to a clean plastic tube.RNAse (100 µl of a 10 mg/ml solution) was added immediately after these extractions and the solution subsequently incubated at 37°C for 30 min.DNA pellets were obtained by adding 0.8 vol. of isopropanol.After washing with 70% ethanol, the DNA pellet was vacuum dried and dissolved in 200 µl of TE buffer (10 mM Tris-HCl, pH 8.0, containing 1 mM EDTA).The quality and concentration of the DNA fragments were evaluated by electrophoresis in 0.8% agarose gels.

Data analysis
Only clearly amplified polymorphic fragments were analyzed.Scores of 1 (present) or 0 (absent) were used to form a matrix.Simple matching coefficients (Sokal and Michener, 1958) were obtained to perform cluster (UPGMA) and principal coordinate analyses (PCoA).Canonical correlations (Hotelling, 1935) were used to investigate the relationship between two variable sets.The first set was represented by PCoA scores (90% total variation) based on RAPD markers, and the second set used the average temperature and humidity values from the geographical zones of the cassava's origin.NTSYS-pc software (Rohlf, 1993) was used for the calculations.Nei's (1973) index of genetic diversity (Ho) was calculated for the four cassava cultivar groups based on allele frequencies of dominant RAPD "loci":

RESULTS
Twenty-one of the 100 primers tested were selected for their quality, quantity and reproducibility capacity of fragments amplified.Eighty-eight of the 193 bands observed were polymorphic (Table II).The average number of fragments amplified per primer and their polymorphism rate were 9.5 (ranging from 6 to 13) and 3.8 (ranging from 2 to 8), respectively.Size of amplified fragments varied from 300 to 2000 pb with an average of 900 pb.An example of PCR amplification is shown in Figure 1.Only the polymorphic bright bands that were clearly amplified were analyzed and the faint bands were not scored.
Genetic similarity coefficients were calculated between all samples taken two by two.Values varied between 0.99 and 0.45, with an average of 0.67.No cultivars had similar profiles.Similar minimal genetic similarity values have been obtained using RAPD markers with other allogamous plants, such as 0.49 for cocoa clones (Wilde et al., 1992), 0.57 for tea (Wachira et al., 1995) and 0.45 for Picea sitchensis (Van de Ven and McNicol, 1995).Furthermore, the minimal similarity value observed among cassava cultivars of this study was greater than that of Marmey et al. (1994), which was 0.64 between the two most divergent African cassava clones.
Similarity coefficients of the four cultivar groups (Santa Isabel, São Paulo, Brazil and the World) were determined (Table III).The Santa Isabel group was significantly different from the other groups.The World group A dendrogram of genetic relationships was established from the similarity coefficient matrix (Figure 2).The hierarchical agglomeration of cultivars was constructed according to the UPGMA method.No strong structure was found among major groups.Nevertheless, most genotypes regrouped on three main branches (A, B and C), or a fourth branch (D), consisting of four samples (Tai1, Mal48, Mal2 and BGM739).All local cultivars from the Amazon (E) belonged to branch A. Cultivars from the State of São Paulo (F), all of Brazil (SRT and BGM) as well as those from the CIAT core collection were well distributed on nearly all branches of the dendrogram.The first two axes of the principal coordinate analysis (PCoA) (Figure 3) were used to structure the four geographically different cassava groups.The percentage of total inertia of axes 1 and 2 was about 14%.Distribution of three of the four cassava groups (São Paulo, Brazil and World) overlapped.Neither axis 1 nor 2 was capable of distinguishing these groups.However, the Santa Isabel group was defined by axis 1.
Shannon genetic diversity indices (Ho) were calculated with all RAPD markers (Table IV).A comparison of the averages revealed significantly less genetic diversity in the Santa Isabel cultivar that the other groups.World collection cultivars were the most diverse (0.93), followed by the State of São Paulo and Brazil, both with 0.91.However, when these differences were examined by comparing averages, they were not statistically significant.Canonical analysis (Figure 4) was performed on 90 of the 126 cassava genotypes that had climatic data available (Table I).In addition, factors calculated by PCoA were based on 88 RAPD markers.The first discriminant function was statistically significant by itself (R = 0.82, P < 0.0052).The two variable groups studied, climatic data and PCoA factors, were statistically correlated.Root coefficients were -0.25 and 0.92 for the variables precipitation and temperature, respectively.
Discriminant analysis was performed on the precipitation data of 90 cassava cultivars (Figure 5).Each variable was correlated with PCoA factors obtained with RAPD markers.Cultivars distributed by the two discriminant functions structured around the variable precipitation.Group P1 consisted of cultivars from weak precipitation zones, group P2 cultivars from average precipita-  tion zones and group P3 cultivars from strong precipitation zones.Group P1 was composed of cultivars from northeastern Brazil, the most arid zone in the country, as well as cultivar ECU 41, the cultivar from the area with the least precipitation among the 90 cultivars included in this study.Group P3 was primarily formed by cultivars from northern Brazil, the Brazilian coast and Central American.Group P2 united cultivars from average precipitation zones, primarily central and southern Brazil.
Discriminant analysis was also performed on the variable temperature (Figure 6).As in the previous analysis, cultivars structured around the variable temperature.All groups, except groups T1 (< 19.9°C) and T2 (20°C-22.9°C),were significantly different (P = 0.05).All groups except the mainly equatorial group T4, were composed of cultivars from several different regions.

DISCUSSION
Cassava has high genetic diversity, which indicates a large genetic base.However, most of the polymorphic RAPD markers used in this study did not reveal much about cassava's genetic structure.It can be supposed that most RAPD markers were well distributed throughout the entire genome.Moreover, these markers revealed an important degree of homoplasy for the structuring of cassava, according to their degree of relationship.These results illustrate that RAPD markers could be useful for structuring the genetic diversity of collections, assessment or formation of a core collection, and especially construction of a genetic map.In addition, these markers are considered valid for identification or characterization of cultivars, establishment of genetic distance between genotypes or identification of identical lines in a collection of clones, as shown in Colombo et al. (1998).
Diversity of local cultivars from the State of São Paulo was identical to that of accessions from the whole country.This result is interesting because São Paulo represents only 1/32 of Brazilian territory.Cassava cultivars collected in this state came from small producers whose produce was destined for domestic consumption.Normally these cultivars have distinct agronomic features based on the producer's preference, which can explain the great diversity found among these cultivars.
Diversity among cultivars distributed throughout Brazil (Brazil group) was also just a little less than acces-   sions representing the world collection, which was composed of American accessions.This result supports the hypothesis that Brazil being is a center of cassava diversity, which corroborates Nassar's report (1978).It has continental dimensions with diverse ecological condi-tions and is historically the largest cassava producer and consumer worldwide.
The large degree of genetic diversity of ethnocultivars from Santa Isabel was surprising.This considerable genetic diversity could be explained by the Indian agricul- tural practice of mixing several cultivars in the same field (Chernela, 1987;Kerr, 1987), possible recombinations between the different cultivars planted, and crosses with wild species.Jennings (1963) showed that in the traditional African cassava culture system plants are sexually reproduced by spontaneous germination of seeds in parcels and cloned to generate new genotypes.This phenomenon has been reported in Brazil (Boster, 1984;Kerr, 1987, Martins, 1994).In plants like cassava, the ease of obtaining plants from seed and the ease of vegetative propagation probably accelerated the process of domestication and culture, which consequently encouraged ample variation of morphological features, root quality, etc.This has also been observed in grapes (Alleweldt et al., 1990).Nevertheless, the absence of strong structuring among cultivars from the four geographical groups of this study, with the exception of the Santa Isabel ethnocultivar group, may be explained by transplanting of cultivars as well as genetic recombinations at different sites.
Other factors, such as human migration, can explain the abundance of genetic diversity in other continents.For example, we estimated that half of the 33 million inhabitants of the State of São Paulo immigrated, mostly from the northeastern and central regions of Brazil.These regions are known as the two main centers of genetic diversity of cassava (Nassar, 1978).Part of these migrants were farmers.Cassava was their basic food, and was taken with them when they moved.The diversity revealed by the Santa Isabel genotypes suggests a certain geographical isolation and limitation of crossing with other cassava cultivars.
The reduced genetic structure observed in relation to the geographical origin of cassava cultivars corroborates the major effects of migration and recombination.As for other vegetatively multiplied species (yam, sweet potato, sugarcane), exchanges of botanical material among producers, which are proportional to the extension of the culture, makes it possible for the same cultivar to be found at several different geographical sites.This is the case with cassava.No one knows the place of origin of this cultivar.A survey of cassava diversity in Africa (Léfèvre, 1993) also revealed little genetic structure in this continent.Léfèvre believed structuring to be associated with the origin of its introduction to the continent, in other words, founder effects.The most recent extension of the cassava culture (beginning about 150 years ago) decreased isolation between parcels and therefore encouraged the flow of genetic material among the different regions.
Nevertheless, cassava structure is associated with its original climate.In this study, cassava groups were differentiated by the precipitation and temperature of their places of origin.Using the variable precipitation, one can see that RAPD markers can separate genotypes based on their adaptability to the availability of water.This tool should help breeders select genotypes appropriate for this type of climate, as suggested by Fukuda et al. (1993) for the semi-arid northeastern region of Brazil.
For the variable temperature, cassava cultures struc- tured between 30°N and 30°S.Cordeiro et al. (1995) also observed a climatic adaptation of clones, while forming a core collection as well as assessing the botanical data of 300 cassava clones.According to Carter et al. (1992), African cassava diffused after its introduction to this vast humid tropical zone in the 16th century, because cultivars were well adapted to these specific climatic conditions.Fukuda (1996) stated that the adaptability of the same cassava genotype to a range of different climatic conditions was not frequent, which would partly explain the wide range of genetic diversity of this species.This study found through RAPD markers that cassava's genetic diversity is great, weakly structured, and tends to structure according to ecological conditions.The manner in which this diversity is distributed in the American continent could be useful for collections and/or conservation of the genetic resources of cassava as well as for breeders.The surprising genetic diversity observed among ethnocultivars from Santa Isabel suggests that the importance of analyzing other peasant communities should be studied.

ACKNOWLEDGMENTS
The authors gratefully acknowledge CAPES for financial support and fellowships conceded to Carlos Colombo.Publication supported by FAPESP.

Figure 1 -
Figure 1 -Example of cassava RAPD amplification (primer N20 Operon) in agarose gel 1.8%.Only the clearly amplified polymorphic bright bands were analyzed (arrows signal the length of these fragments in pairs of bases).The faint bands were not scored.
Figure 2 -Dendrogram based on UPGMA model calculated from genetic distances among American cassava genotypes from Santa Isabel (A), São Paulo (B), Brazil (C) and world collection (D), employing 88 RAPD markers.

Figure 3 -
Figure 3 -Plot defined by the first two principal coordinate analysis factors based on genetic similarities (simple matching coefficients) of 126 cassava genotypes.Similarity coefficients were calculated with 88 RAPD markers.

Figure 4 -
Figure 4 -Canonical analysis of climatic data (temperature and precipitation) from 90 cassava cultivar sites and principal coordinate analysis (PCoA) factors, calculated with 88 RAPD markers.

Figure 5 -
Figure 5 -Scattering of 90 cassava cultivars on a plot defined by discriminant function 1, based on precipitation (P1 to P3) of cultivar's place of origin, and discriminant function 2, based on the first 40 principal coordinate analysis (PcoA) factors.*,**Statistically significant at 5% and 1%, respectively.

Figure 6 -
Figure 6 -Scattering of 90 cassava cultivars on a plot defined by discriminant function 1, based on temperature (T1 to T3) of cultivar's place of origin, and discriminant function 2, based on the first 40 principal coordinate analysis (PcoA) factors.ns, Nonsignificant.*Statistically significant at 5%.

Table I -
Plant samples used to investigate the genetic relatedness among cassava cultivars.
* Cassava germplasm from Santa Isabel (A), São Paulo (B), Brazil (C) and World collection (D) and local average precipitation (Precip.) and temperature (Temp.).Table I -Continuedwas the most diverse.Surprisingly, similar results were found for the cultivars from the State of São Paulo.Local cultivars collected in this state were almost as diverse as those of the world collection.

Table II -
Number and frequency of RAPD polymorphism products amplified by the primer in cassava.

Table III -
Simple matching coefficient values for cassava cultivar groups: Santa Isabel local cultivars, São Paulo State cultivars, accessions of the Brazilian group and accessions of the World collection.

Table IV -
Shannon genetic index for the four cassava cultivar groups and comparisons of their averages by t-test (Student-Fisher).