Genetic variability and heterozygosity of cocoa accessions of Parinari (Pa) population, based on microsatellite markers

The objective of this work was to study the genetic variability and the heterozygosity of accessions of the Parinari (Pa) series in the collection of the Active Germplasm Bank (BAG) of the Cocoa Research Center (CEPEC), CEPLAC, Bahia using microsatellite markers. In the present study, 34 accessions of Theobroma cacao L. were used, being 25 of the Parinari series. Twelve microsatellite loci were analyzed generating a total of 49 alleles. Results indicated genetic variability among the Parinari accessions with genetic distances varying from 0.00 to 0.64 and heterozygosity levels between 11 and 72%. About 40% of the accessions presented heterozygosity equal or greater than 50%.


INTRODUCTION
Cocoa (Theobroma cacao L.) belongs to the Sterculiaceae family, which contains about 50 genus, the most important being Theobroma. Based on seed and fruit characteristics and the geographical distribution of cocoa, two racial groups are known (Criollo and Forastero). A third group (Trinitário), probably originated from the hybridization between Forastero and Criollo. Forasteros can be further subdivided into Upper and Lower Amazon, representing genetic materials responsible for 80% of the world cocoa production (Lerceteau et al., 1992;Marita, 1998).
The Parinari accessions were collected in the Parinari River area, in the Peruvian Amazon (Pound, 1938). These accessions were collected with the objective of selecting resistance sources to the witches´ broom Sounigo et al., KEY WORDS: Theobroma cacao L, germplasm, genetic diversity, molecular markers. 2001). Although reports exists of resistance to witches´ broom in some clones of this series (Bartley, 1983;Umaharan et al., 2001), such resistance has not been confirmed in the Active Germplasm Bank (BAG), of the Cocoa Research Center (CEPEC). However, resistance to Phytophthora sp., in fruits, has been observed in many of these clones (Iwaro et al., 2001;Luz et al., 1996;Pires et al., 1994). The accessions Pa 30 and Pa 150 were some of the few that presented resistance to the three main Phytophthora species (Luz et al., 1996). Also, under field conditions, Pa 300 presented low incidence to Phytophthora spp. (Pires et al., 1994).
Isoenzymatic and compatibility studies suggest that few trees contributed to the formation of the Parinari series, indicating little variability among accessions of this series (Yamada and Guries, 1994;. In these studies some alternative isoenzymatic alleles were not found and few alleles of incompatibility were observed within the population. It is suggested that seven trees contributed to the formation of the Parinari population (Lockwood and End, 1993).  evaluated the genetic diversity of 189 clones of 14 populations and 89 clones of 11 populations using isoenzymes and RAPD respectively, and showed a high degree of genetic diversity within the Scavina population and low Crop Breeding and Applied Biotechnology, v. 3, n. 4, p. 289-296, 2003 within P and IMC. The Parinari accessions presented low genetic diversity with isoenzymes and high diversity using RAPD Sankar and Sounigo, 1995;. In the grouping analysis, P and Pa clustered together, being different from the other two groups, the Sca and the IMC accessions. Bekele and Bekele (1996), using 28 quantitative and 26 qualitative morphologic descritors and 100 accessions of 24 populations, showed that among the 11 groups formed, Pa accessions were represented in 3 different groups. Charters and Wilkinson (2000), using ISSR markers, clustered the 62 cocoa accessions studied in 14 groups, with the three accessions of Parinari series in the same group. In a more extensive study, Marita (1998) evaluating 270 accessions using RAPD demonstrated that the seven Parinari accessions clustered closely, except Pa 175.
The objective of this work was to study the genetic variability and heterozygosity of Parinari accessions from the germplasm collection at the CEPEC/ CEPLAC based on microsatellite markers.

Genetic material
In the present study 34 accessions of Theobroma cacao L were used, being 25 of the Parinari (Pa) series with two of them repeated in different rows of the collection, and with 7 suspects to be of the same geographical origin (Table 1).

DNA extraction
Leaves of each accession of T.cacao were collected for DNA extraction. The genomic DNA of each accession was extracted using the CTAB method (Doyle and Doyle, 1990), with some modifications (Araújo et al., 2000). After the extraction, the DNA concentration was estimated by spectrophotometer at 260 nm (Sambrook et al., 1989). Bands of total genomic DNA separated by electrophoresis in agarose gel of 0.8% were used as indicative of the integrity and purity of the extracted DNA. After quantification, the DNA samples of good quality were diluted to 10ng/mL.

Microsatellite markers
The amplification reactions for microsatellite were prepared in a total volume of 15 ml, including Tris-HCl 10 mM (pH 8.3), KCl 50 mM, MgCl2 2.4 mM, 150 mM for each one of the desoxinucleotides (dATP, dTTP, dGTP and dCTP), 3 pM for each one of the two primers (F and R), a unit of the enzyme Taq polymerase and approximately 30 ng of DNA. The amplifications were made in thermocycler, according to the following program: 4 minutes at 94 ºC + 10 cycles (30 seconds at 94 ºC + 60 seconds at 60 ºC -1 ºC to each cycle + 90 seconds at 72 ºC) + 30 cycles (30 seconds at 94 ºC + 60 seconds at 48 ºC + 90 seconds at 72 ºC) + 6 minutes for 72 ºC. After amplification, the temperature of the samples was reduced to 4ºC. Twelve microsatellite pairs of primers were used (Table 2).
For each sample, 3 ml of a mixture of bromophenol blue (0.25%), glycerol (60%) and water (39.75%), were added. These samples were loaded in agarose gel 3% for separation of microsatellite fragments. The gel was submerged in TBE buffer (Tris-borate 90 mM, EDTA 1 mM) and electrophoretic separation was conducted for approximately four hours, at 100 volts. At the end of the run, the gels were stained with ethidium bromide and photographed under ultraviolet light.

Statistical analyses
The microsatellite markers generated were transformed to a coded numeric matrix, from which genetic distances were calculated between the genetic materials and cluster analyses done. The code of the molecular markers were 0 for absence of the allele, 1 for presence of one copy of the allele and 2 for presence of two copies of the allele.
The genetic distances were calculated according to the following formula: DGij = 1 -((ΣPLCij) / NTLij) being: DGij = genetic distance between the accessions i and j; PLCij = proportions of coincidents alleles in each loci, being value 1 for 2 coincidents allele (in homozygous or heterozygous); 0.5 for one coincident allele and zero for nil coincident allele. NTL = total number of loci without missing data to I or j.
Defined distance matrix, the multidimensional space Crop Breeding and Applied Biotechnology, v. 3, n. 4, p. 289-296, 2003 was reduced to bidimensional by the 'multidimensional scaling´ method-MDS SAS and cluster analyses were done using the UPGMA method considering two groups (SAS, 1990).
The heterozygosity level was estimated as the ratio between the number of heterozygous locus and the total number of analyzed loci.

RESULTS AND DISCUSSION
The twelve microsatellite loci analyzed generated a total of 49 alleles, with an average of 4 alleles per locus ( Table 2) and (Figure 1). Considering only Parinari accessions, this number drops to 44 alelles, with an average of 3.7 alelles per locus. Faleiro et al. (2001a) analyzing microsatellite loci in Upper Amazon accessions obtained an average of 7.5 alleles per locus. Risterucci et al. (2000), also analyzing microsatellites in cocoa, obtained an average of 5.6 alleles per locus. The low number of alleles per locus in our work is an indication of a certain genetic relationship among the accessions of the Parinari series.
The grouping analysis showed a certain relationship among the accessions of the Parinari series ( Figure  2). The clones from 1 to 6, suspected of being Parinari, formed one group and the remaining known as being of the Parinari series formed another group.
The information about the origin of the Parinari accessions is scarce, but they seem to be originated from few trees, as indicated by Lockwood and End (1993). Most of the Parinari accessions should be genetically related as half-sibs or full -sibs, evidenced by the low number of incompatibility alleles .
In isoenzyme studies, the alternative forms of MDH and IDH alleles (Yamada and Guries, 1994) were not observed, suggesting that the Parinari populations collected were not in Hardy-Weinberg equilibrium. In cocoa, pollination by insects is common, resulting in a predominance of full-sib families (Yamada and Guries, 1998). The absence of alternative forms of isoenzymes alleles and the presence of few compatibility alleles suggest that a small number of parents participated in the formation of that population. Also, the low number of microsatellite alleles per locus within this population is an indication of a low number of parents. Marita (1998) also found Pa clones very closely related in cluster analysis of 270 clones. Similar results were found in the case of Pa 56, Pa 88 and Pa 184, in studies with other cocoa accessions (Charters and Wilkinson, 2000).
The genetic materials 1 and 2 (accession 192) were collected in different rows of the germplasm collection, but they seem to be the same material (Figure 2 and genetic similarity 1.00, data not shown). CEPEC accessions (526, 527, 528 and 529) were clustered and are quite related among themselves. Those four CEPEC accessions were grouped and seem to be different from TSA 644 whose row in BAG/CEPEC was initially collected. These four accessions are different and do not seem to be related to Parinari accessions, as initially believed. The  Table 1. Average heterozygosity of cocoa accessions based on 12 microsatellites loci.
genetic materials 8 and 34 are classified as Pa 107 , however, the results demonstrate that in fact they are different materials (genetic similarity 0.55, data not shown). Mistakes and mixtures in germplasm banks are not rare and deserve special attention by researchers (Figueira, 1998;Faleiro et al., 2001b;Sounigo et al., 2001;Saunders et al., 2001).
In spite of the fact that Parinari clones are related to each other, variability exists for morphology and size of the fruits, which can be explained by the divergence and heterozygosity of the parents that formed the Parinari population, making the progenies to segregate for those characteristics. In these cases, it is possible that the parents' heterozygosity resulted in high variability of the progenies that is confirmed with heterozygosity data of the Parinari accessions ( Table 1) that varied from 11% to 72%, being equal or more than 50% in 40% of the accessions. The average heterozygosity was 44.2%, which is higher than the 23.7% found by Crouzillat et al. (2000).  107 showed the highest levels.
The genetic variability of Parinari accessions is also confirmed by the genetic distances that varied from 0.00 to 0.64 (data not shown). Christhoper and  found an average similarity index of 0.66 using RAPD. This data indicates a high genetic diversity within Parinari populations. Also, using morphological qualitative and quantitative descriptors, Parinari accessions were clustered into 3 different groups (Bekele and Bekele, 1996). Using isoenzymes, variability was not found, probably because the low number of enzyme systems used .
The accessions of Parinari series has a certain genetic relationship. However, data of genetics distance and heterozygosity obtained in this work demonstrate the existence of genetic variability.