**4. Characterization and evaluation of** *Hevea* **germplasm**

In order to use the germplasm in breeding programs, it must be characterised and evaluated. There is often a delay between collection of germplasm and its evaluation, particularly for rubber trees because of the time required for them to reach maturity. Evaluation is useful if it considers the traits wanted by plant breeders. We are fortunate that our program of germplasm acquisition and evaluation is very closely linked to our program of *Hevea* breeding with the same people usually involving in both. Preliminary evaluation can help indicate those accessions that need more detailed evaluation, but those that appear not to be of immediate values should not be simply discarded.

#### **4.1 Agronomical characteristics**

Standard characterization and evaluation of germplasm collection may be routinely performed using different methods including traditional practices such as the use of descriptive lists of morphological characters. They may also involve evaluation of the agronomical performances under various environmental conditions. Understanding the nature and the magnitude of variability of important traits existing among plant genetic materials is vital for the effective utilization of such materials for breeding purposes. In rubber tree, high latex yield is always the exclusive objective of breeding programs. To this ultimate objective, many different factors are associated. For instance, the main components of productivity are the growth of the trunk determined during the immature period before the beginning of tapping, the resistance to various diseases and the tolerance to stress factors such as high altitude, low temperature, wind damage and moisture deficit.

In Vietnam, a part of *Hevea* germplasm, especially IRRDB'81 collection, has been agronomically evaluated in various clonal trials including arboreta and small scale clonal trials at different locations. In the view of latex production, IRRDB'81 collection exhibited very poor performance with an average latex yield of around 16% of the level of the currently developed Wickham clones after 5 years of tapping. This result was similar to that of other studies conducted on IRRDB'81 collections in Malaysia, Indonesia, Ivory Coast and China (Ramli et al., 2004; Aidi et al., 2002; Clément-Demange et al., 2002; Hu et al., 2002). Outstandingly, in the first three years of tapping, some IRRDB'81 accessions such as AC56/276, AC62/54, MT8/27, MT/I/2 and RO62/26 produced 30.0 – 45.0 gram/tree/tapping, or 102.7 – 153.8% of the production of the control clone (GT1). These accessions have been being used as parents in *Hevea* breeding programs. Considering geographical origins, accessions derived from Mato Grosso seemed to be better yielder than those from Acre and Rondonia. This might indicate their better adaptability to the experimental areas where the climate featured by a distinct dry spell of six months is similar to that of the original region which is known to have a dry spell of three to four months annually (Chevallier, 1988). A large number of IRRDB'81 accessions showed very good growth performance with girth at opening ranging from 59.9 – 74.0 cm, or 124.8 – 133.5% of that of the control clone. Remarkably, some IRRDB'81 accessions such as MT29/68, RO24/58 and RO32/104 had significantly higher girth than the control when planted in the highland area which was considered as a non-traditional rubber region in Vietnam. The superiority of these accessions will be of considerable value in advanced breeding programs.

Recently, IRRDB'81 collection has been considered as an important source for timber selection and rubber wood production. The average wood volume of the IRRDB'81

In order to use the germplasm in breeding programs, it must be characterised and evaluated. There is often a delay between collection of germplasm and its evaluation, particularly for rubber trees because of the time required for them to reach maturity. Evaluation is useful if it considers the traits wanted by plant breeders. We are fortunate that our program of germplasm acquisition and evaluation is very closely linked to our program of *Hevea* breeding with the same people usually involving in both. Preliminary evaluation can help indicate those accessions that need more detailed evaluation, but those that appear

Standard characterization and evaluation of germplasm collection may be routinely performed using different methods including traditional practices such as the use of descriptive lists of morphological characters. They may also involve evaluation of the agronomical performances under various environmental conditions. Understanding the nature and the magnitude of variability of important traits existing among plant genetic materials is vital for the effective utilization of such materials for breeding purposes. In rubber tree, high latex yield is always the exclusive objective of breeding programs. To this ultimate objective, many different factors are associated. For instance, the main components of productivity are the growth of the trunk determined during the immature period before the beginning of tapping, the resistance to various diseases and the tolerance to stress factors

In Vietnam, a part of *Hevea* germplasm, especially IRRDB'81 collection, has been agronomically evaluated in various clonal trials including arboreta and small scale clonal trials at different locations. In the view of latex production, IRRDB'81 collection exhibited very poor performance with an average latex yield of around 16% of the level of the currently developed Wickham clones after 5 years of tapping. This result was similar to that of other studies conducted on IRRDB'81 collections in Malaysia, Indonesia, Ivory Coast and China (Ramli et al., 2004; Aidi et al., 2002; Clément-Demange et al., 2002; Hu et al., 2002). Outstandingly, in the first three years of tapping, some IRRDB'81 accessions such as AC56/276, AC62/54, MT8/27, MT/I/2 and RO62/26 produced 30.0 – 45.0 gram/tree/tapping, or 102.7 – 153.8% of the production of the control clone (GT1). These accessions have been being used as parents in *Hevea* breeding programs. Considering geographical origins, accessions derived from Mato Grosso seemed to be better yielder than those from Acre and Rondonia. This might indicate their better adaptability to the experimental areas where the climate featured by a distinct dry spell of six months is similar to that of the original region which is known to have a dry spell of three to four months annually (Chevallier, 1988). A large number of IRRDB'81 accessions showed very good growth performance with girth at opening ranging from 59.9 – 74.0 cm, or 124.8 – 133.5% of that of the control clone. Remarkably, some IRRDB'81 accessions such as MT29/68, RO24/58 and RO32/104 had significantly higher girth than the control when planted in the highland area which was considered as a non-traditional rubber region in Vietnam. The superiority of these accessions will be of considerable value in advanced breeding programs. Recently, IRRDB'81 collection has been considered as an important source for timber selection and rubber wood production. The average wood volume of the IRRDB'81

**4. Characterization and evaluation of** *Hevea* **germplasm** 

not to be of immediate values should not be simply discarded.

such as high altitude, low temperature, wind damage and moisture deficit.

**4.1 Agronomical characteristics** 

collection conserved in Vietnam was initially recorded, yielding 1.48 m3/tree at the age of 21 years, 43.4% higher than the Wickham population. Among these, several accessions appeared to be the best with clear bole volume ranging from 1.55 to 2.07 m3/tree, which could be considered as suitable clones for timber production purposes. Similarly, several high timber yield accessions of the IRRDB'81 collection in Indonesia and Malaysia were also reported. At the age of 13 years, a total of 28 accessions in Indonesia and 20 accessions in Malaysia were selected for timber yield with clear bole volume at a range of 0.90 to 2.56 m3/tree (Aidi et al., 2002) and 1.0 to 1.6 m3/tree (Ramli et al., 2004), respectively.

It is known that unlike other clonally multiplied species, *Hevea* is not affected by viral diseases (Simmonds, 1989). Other diseases which are considered as economic importance are *Gloeosporium* leaf disease (*Colletotrichum gloeosporioides* Pen. Sacc.), pink disease (*Corticium salmonicolor* Berk. & Br.), powdery mildew (*Oidium heveae* Stein.), *Corynespora* leaf fall (*Corynespora cassiicola* Berk. & Curt. Wei.), *Phytophthora* leaf fall (*Phytophthora* sp.) and SALB (South American Leaf Blight - *Microcyclus ulei* P. Henn von Arx.). Among these diseases, *Corynespora* leaf fall and SALB are the most important in rubber plantations. Wycherly (1969) noted that the clonal and location specificity was evident towards resistance to these diseases. Differences in the level of resistance to important diseases have been observed in IRRDB'81 collections from different countries. In India, a large number of IRRDB'81 accessions were showed to be resistant to powdery mildew (140 accessions) and *Corynespora* leaf fall (70 accessions) (Varghese et al., 2002). Similarly, a total of 21 accessions in IRRDB'81 collection are resistant to powdery mildew in bush-wood garden in China (Huang et al., 2002). The resistance to SALB was observed on 298 accessions from *ex situ* germplasm collection planted in both French Guyana and Brazil, of which the accessions from Acre and Rondonia were the most resistant (Le Guen et al., 2002). In Vietnam, it seemed that Wickham and IRRDB'81 accessions showed no significant difference in susceptibility to pink disease, powdery mildew and *Gloeosporium* leaf diseases (Lam et al., 2002). In general, evaluations of IRRDB'81 collection for biotic and abiotic stresses are in progress in major rubber growing countries.

#### **4.2 Genetic diversity based on biochemical and molecular markers**

In conventional plant breeding, many morphological traits have been used as markers for genetic analyses and cultivar identification, but specific genetic information on Mendelian traits are rare in *Hevea*. In contrast to traditional practices, genetic characterization refers to the description of the attributes that follow Mendelian inheritance or involve specific DNA sequences. In this way, biochemical assays, which detect differences between isozymes, or molecular markers were applied to determine the genetic diversity of the germplasm. In addition, the development of molecular and biochemical markers help researchers not only to identify genotypes, but also to assess and exploit the genetic variability (Whitkus et al., 1994). Insights into the relative genetic diversity among *Hevea* collections would be useful in *Hevea* breeding as well as *ex situ* conservation of *Hevea* genetic resources. The commercial value associated with identifying useful traits, especially yield and growth, would create a direct value in genebanks, ensuring long-term preservation of a collection. Moreover, *Hevea* germplasm characterization using molecular and biochemical markers will contribute to the knowledge of genetic relationships not only among wild accessions but also between accessions of wild and cultivated gene pool, and hence help to facilitate the breeding programs. In Vietnam, significant progress has been made in evaluation and characterization of *Hevea* germplasm by applying genetic markers. Among several efficient methods to reveal the genetic variability within and among plant populations, the most widely applied methods are isozyme electrophoresis and random amplified DNA polymorphism (RAPD). Both markers are useful to analyze genetic diversity of *Hevea* germpasm, and to select good *Hevea* clones for future breeding or cultivation purposes.

#### **4.2.1 Isozymes marker**

As first described by Hunter and Markert (1957), isozymes were defined as the different molecular forms in which proteins may exist with the same enzymatic specificity (Buth, 1984). This means that different variants on the same enzymes have identical or similar functions and are present in the same individual. Isozyme had played a minor role in research on plant biochemistry until genetic polymorphism for isozymes within the same population was discovered in 1966 (Stebbins, 1989; Wendel, 1989). In the 1980s, analysis of isozymes was developed at CIRAD (French Agricultural Research Centre for International Development) with 13 polymorphic isozymic systems to formulate a diagnostic kit associated with a clonal identification database. This kit has proved to be able to differentiate a large set of cultivated clones (Leconte et al., 1994). Since then, isozymes have been used as genetic markers for identification of rubber tree cultivars, genetic diversity analysis, controlling progenies issued from hand pollination and reproductive biology (Chevallier, 1988; Leconte et al., 1994; Paiva et al., 1994; Sunderasan et al., 1994). The first study on using isozyme markers for a precise understanding of the genetic diversity of the wild *Hevea* germplasm was carried out using ten isozyme markers on a set of 263 accessions from the IRRDB'81 collection (Chevallier, 1988).

Isozyme markers were firstly used in Vietnam to identify and confirm rubber clones in budwood gardens. Moreover, it was also used in *Hevea* breeding for hybrid genealogical legitimacy and genetic diversity research of *Hevea* germplasm. A total of 12 isozyme systems were used in studying genetic diversity of the IRRDB'81 *Hevea* germplasm in Vietnam. Banding patterns of representative IRRDB'81 accessions generated by isozyme electrophoresis are shown in Figure 4.

The study was performed on both IRRDB'81 collection and Wickham population with 117 accessions from 15 districts of the states of Acre, Rondonia and Mato Grosso of Brazil and 24 Wickham's clones (Fig. 5 and Table 2). The quantity of accessions sampled for each district was more or less proportional to the quantity of accessions currently conserved for the district. The result of isozymatic analysis showed that a total of 60 alleles were detected (Table 2). Out of them, 60 alleles were detected in IRRDB'81 accessions and 26 alleles in Wickham population. The result showed that the alleles detected in Wickham population were also detected in the IRRDB'81 collection, many new alleles were found in the IRRDB'81 collection only, thus underlining the genetic enrichment provided by the wild *Hevea* collections. Among IRRDB'81 collection, Acre accessions had the largest number of polymorphic alleles (51/60 alleles), followed by the Rodonia (45/60 alleles) and Mato Grosso accessions (45/60 alleles). This result revealed that Mato Grosso was obviously less polymorphic than Rondonia because Mato Grosso and Rondonia groups had the same number of detected polymorphic alleles but the volume of samples of the Mato Grosso groups was bigger than that of the Rondonia groups. The result also showed the significant polymorphic differences among the districts. The number of alleles detected in the districts Brasileia, Taurauca of Acre and district Cartriquacu of Mato Grosso was high in spite of the

characterization of *Hevea* germplasm by applying genetic markers. Among several efficient methods to reveal the genetic variability within and among plant populations, the most widely applied methods are isozyme electrophoresis and random amplified DNA polymorphism (RAPD). Both markers are useful to analyze genetic diversity of *Hevea* germpasm, and to select good *Hevea* clones for future breeding or cultivation purposes.

As first described by Hunter and Markert (1957), isozymes were defined as the different molecular forms in which proteins may exist with the same enzymatic specificity (Buth, 1984). This means that different variants on the same enzymes have identical or similar functions and are present in the same individual. Isozyme had played a minor role in research on plant biochemistry until genetic polymorphism for isozymes within the same population was discovered in 1966 (Stebbins, 1989; Wendel, 1989). In the 1980s, analysis of isozymes was developed at CIRAD (French Agricultural Research Centre for International Development) with 13 polymorphic isozymic systems to formulate a diagnostic kit associated with a clonal identification database. This kit has proved to be able to differentiate a large set of cultivated clones (Leconte et al., 1994). Since then, isozymes have been used as genetic markers for identification of rubber tree cultivars, genetic diversity analysis, controlling progenies issued from hand pollination and reproductive biology (Chevallier, 1988; Leconte et al., 1994; Paiva et al., 1994; Sunderasan et al., 1994). The first study on using isozyme markers for a precise understanding of the genetic diversity of the wild *Hevea* germplasm was carried out using ten isozyme markers on a set of 263 accessions

Isozyme markers were firstly used in Vietnam to identify and confirm rubber clones in budwood gardens. Moreover, it was also used in *Hevea* breeding for hybrid genealogical legitimacy and genetic diversity research of *Hevea* germplasm. A total of 12 isozyme systems were used in studying genetic diversity of the IRRDB'81 *Hevea* germplasm in Vietnam. Banding patterns of representative IRRDB'81 accessions generated by isozyme

The study was performed on both IRRDB'81 collection and Wickham population with 117 accessions from 15 districts of the states of Acre, Rondonia and Mato Grosso of Brazil and 24 Wickham's clones (Fig. 5 and Table 2). The quantity of accessions sampled for each district was more or less proportional to the quantity of accessions currently conserved for the district. The result of isozymatic analysis showed that a total of 60 alleles were detected (Table 2). Out of them, 60 alleles were detected in IRRDB'81 accessions and 26 alleles in Wickham population. The result showed that the alleles detected in Wickham population were also detected in the IRRDB'81 collection, many new alleles were found in the IRRDB'81 collection only, thus underlining the genetic enrichment provided by the wild *Hevea* collections. Among IRRDB'81 collection, Acre accessions had the largest number of polymorphic alleles (51/60 alleles), followed by the Rodonia (45/60 alleles) and Mato Grosso accessions (45/60 alleles). This result revealed that Mato Grosso was obviously less polymorphic than Rondonia because Mato Grosso and Rondonia groups had the same number of detected polymorphic alleles but the volume of samples of the Mato Grosso groups was bigger than that of the Rondonia groups. The result also showed the significant polymorphic differences among the districts. The number of alleles detected in the districts Brasileia, Taurauca of Acre and district Cartriquacu of Mato Grosso was high in spite of the

**4.2.1 Isozymes marker** 

from the IRRDB'81 collection (Chevallier, 1988).

electrophoresis are shown in Figure 4.

small volume of the samples whereas the number of alleles detected in districts Itanba and Aracatuba of Mato Grosso was quite low.


Fig. 4. Banding patterns of representative *Hevea* accessions generated by isozyme electrophoresis. Twelve isozyme systems used in studying genetic diversity of the IRRDB'81 *Hevea* germplasm had very different banding patterns. MDH, malate dehydrogenase; PGI, phospho glucose isomerase; AAP, alanyl amino peptidase; LAP, leucine amino peptidase; EST, esterase; DIA, diaphorase; ADH, alcohol dehydrogenase; ICD, isocitrate dehydrogenase; PGD, phosphogluconase dehydrogenase; PGM, phosphoglucomutase; GOT, glutamate oxaloacetate transaminase; SKD, shikimate acid

The study affirmed that the genetic base of *Hevea* germplasm in Vietnam, especially IRRDB'81 collection, was prosperous and diversified. Because of the high genetic variability level, this collection would contribute effectively to the long term progress of *Hevea* breeding and selection program in the country. In contrast, the Wickham population showed a low level of genetic variability, which is the consequence of the oriented selection through many years in a narrow geographical origin.

The result of cluster analysis based on isozymes database revealed the noticeable relationship between genetic clusters (Fig. 6). According to the genetic distance between the accessions, Acre accessions and Rondonia accessions were close to each other, meanwhile the genetic distance between Acre or Rondonia accessions were far from Mato Grosso accessions, except that the accessions from Vila Bella district of Mato Grosso (MT/VB) were close to those of Rondonia. Among IRRDB'81 accessions, Mato Grosso population was relatively close to Wickham collection based on isozyme analysis. In general, the genetic distance between the accessions conformed to the geographical origins of *Hevea*. However, several accessions of Arce and Rondonia were not separated distinctly using isozyme electrophoresis although they were distributed widely according to the geographical origins.

Fig. 5. Geographical origins of *Hevea* IRRDB'81 collection

Fig. 6. Dendrogram based on Nei's genetic distance between IRRDB'81 collection and Wickham population.



Fig. 5. Geographical origins of *Hevea* IRRDB'81 collection

Wickham population.

Fig. 6. Dendrogram based on Nei's genetic distance between IRRDB'81 collection and

\* MDH, malate dehydrogenase; PGI, phospho glucose isomerase; AAP, alanyl amino peptidase; LAP, leucine amino peptidase; EST, esterase; DIA, diaphorase; ADH, alcohol dehydrogenase; ICD, isocitrate dehydrogenase; PGD, phosphogluconase dehydrogenase; PGM, phosphoglucomutase; GOT, glutamate oxaloacetate transaminase; SKD, shikimate acid

Table 2. Genetic variability of *Hevea* germpalsm based on isozymes

The results of isozymes analysis indicated that the *Hevea* germplasm conserved in Vietnam is very diversed. This characterisation would help to utilize the new genetic resources more effectively in *Hevea* breeding programs. The combination of morphological characterization and isozyme markers could help breeders to constitute a core collection of *Hevea* IRRDB'81 germplasm to ensure the conservation of the genetic variability. In addition, maintaining the genetic variability in *Hevea* germplasm would help to reduce gene erosion. Moreover, isozyme markers could be used as an assistant tool to orient a long term plan to advance heterosis to improve Wickham materials based on the recombination between Wickham clones and IRRDB'81 accessions. However, isozyme-based analysis is limited by the rather small number of marker loci available and a general lack of polymorphism for these loci. In addition, the analysis has to be carried out near the field sites owing to the fragility of the isozymes to varied temperatures or otherwise the samples need to be freeze-dried and transported to the laboratory. In spite of such limitations, isozyme was still a helpful marker to evaluate the genetic variability of the *Hevea* germplasm in Vietnam.

#### **4.2.2 RAPD marker**

The random amplified polymorphic DNA (RAPD) technique, first described by Williams et al. (1990), despite some limitations, has provided a useful approach for evaluating population's genetic differentiation, particularly in species that are poorly genetically known (Silva and Russo, 2000; Nybom, 2004). Recently, a large number of studies have pointed out that DNA-based markers, such as RAPD, were superior to isozymes in detecting genetic diversity (Garkava et al., 2000; Matos et al., 2001; Ochiai et al., 2001; Sharma et al., 2008). It is known that isozymes represent allelic expression of the same locus, while DNA fragments produced by RAPD are independent genetic markers (Ochiai et al., 2001) with a lower proportion of non-neutral markers than formerly (Bartish et al., 2000). Hence, isozyme and RAPD analyses often give discordant patterns, suggesting the importance of using multiple molecular marker systems in studies of population structure (Wendel and Doyle, 1998; Bartish et al., 2000; Lebot et al., 2003). RAPD marker was also used to evaluate the levels of gene flow between species (Arnold et al., 1991) and detection of gene introgression in various plant species (Waugh et al., 1992; Orozco-Castillo et al., 1994; Gomez et al., 1996). In rubber tree, RAPD has become a useful maker for investigating genetic diversity within and between *Hevea* populations, especially the IRRDB'81 collection (Varghese et al., 1997; Venkatachalam et al., 2002; Lam et al., 2009). Moreover, this marker was also used to identify a dwarf genome-specific marker (Venkatachalam et al., 2004) or certain homology to proline-specific permease gene (Venkatachalam et al., 2006) in rubber tree. The accumulated data on *Hevea* RAPD analysis from different accessions give information on genetic relations and *Hevea* origin, and provide the initial basis for clonal distinction and germplasm evaluation of agronomical interest. Therefore, the data can also be used in *Hevea* improvement programs.

In Vietnam, RAPD was firstly used to study genetic diversity of *Hevea* germplasm (Lam et al., 2009). The study was performed on IRRDB'81 collection with 59 accessions from 13 districts of the states of Acre, Rondonia and Mato Grosso of Brazil (Fig. 5). Using 6 oligonucleotide primers, the percentage of polymorphic loci calculated for individual districts ranged from 15.38% in Assis-Brasil district to 70.77% in Sena Madureira district of Acre, which totally had 10 and 46 polymorphic banding patterns, respectively (Table 3, Fig. 7). Although the sample sizes might have certain effects on the extent of the polymorphism of various districts, in the cases of Assis-Brasil of Acre, Ariquemes of Rondonia, and Vila Bela of Mato Grosso, they were quite different in the extent of polymorphism with the same sample sizes. In addition, the Jiparana district of Rondonia was very polymorphic regardless of its small sample size.

Fig. 7. DNA fingerprints of representative IRRDB'81 accessions generated by primer A18 (left) and OPB-12 (right)

The random amplified polymorphic DNA (RAPD) technique, first described by Williams et al. (1990), despite some limitations, has provided a useful approach for evaluating population's genetic differentiation, particularly in species that are poorly genetically known (Silva and Russo, 2000; Nybom, 2004). Recently, a large number of studies have pointed out that DNA-based markers, such as RAPD, were superior to isozymes in detecting genetic diversity (Garkava et al., 2000; Matos et al., 2001; Ochiai et al., 2001; Sharma et al., 2008). It is known that isozymes represent allelic expression of the same locus, while DNA fragments produced by RAPD are independent genetic markers (Ochiai et al., 2001) with a lower proportion of non-neutral markers than formerly (Bartish et al., 2000). Hence, isozyme and RAPD analyses often give discordant patterns, suggesting the importance of using multiple molecular marker systems in studies of population structure (Wendel and Doyle, 1998; Bartish et al., 2000; Lebot et al., 2003). RAPD marker was also used to evaluate the levels of gene flow between species (Arnold et al., 1991) and detection of gene introgression in various plant species (Waugh et al., 1992; Orozco-Castillo et al., 1994; Gomez et al., 1996). In rubber tree, RAPD has become a useful maker for investigating genetic diversity within and between *Hevea* populations, especially the IRRDB'81 collection (Varghese et al., 1997; Venkatachalam et al., 2002; Lam et al., 2009). Moreover, this marker was also used to identify a dwarf genome-specific marker (Venkatachalam et al., 2004) or certain homology to proline-specific permease gene (Venkatachalam et al., 2006) in rubber tree. The accumulated data on *Hevea* RAPD analysis from different accessions give information on genetic relations and *Hevea* origin, and provide the initial basis for clonal distinction and germplasm evaluation of agronomical interest. Therefore, the data can also

In Vietnam, RAPD was firstly used to study genetic diversity of *Hevea* germplasm (Lam et al., 2009). The study was performed on IRRDB'81 collection with 59 accessions from 13 districts of the states of Acre, Rondonia and Mato Grosso of Brazil (Fig. 5). Using 6 oligonucleotide primers, the percentage of polymorphic loci calculated for individual districts ranged from 15.38% in Assis-Brasil district to 70.77% in Sena Madureira district of Acre, which totally had 10 and 46 polymorphic banding patterns, respectively (Table 3, Fig. 7). Although the sample sizes might have certain effects on the extent of the polymorphism of various districts, in the cases of Assis-Brasil of Acre, Ariquemes of Rondonia, and Vila Bela of Mato Grosso, they were quite different in the extent of polymorphism with the same sample sizes. In addition, the

Jiparana district of Rondonia was very polymorphic regardless of its small sample size.

Fig. 7. DNA fingerprints of representative IRRDB'81 accessions generated by primer A18

**4.2.2 RAPD marker** 

be used in *Hevea* improvement programs.

(left) and OPB-12 (right)


\* Genetic distance is mean genetic distance among accessions in individual district

Table 3. Patterns of genetic diversity of *Hevea* IRRDB'81 collection based on RAPD analysis

In general, the IRRBD'81 collection conserved in Vietnam showed the high level of genetic diversity detected by RAPD. In fact, the mean values of heterozygosity or Nei's genetic diversity (Nei, 1978) within individual districts varied from 0.064 to 0.264 over 65 loci with the average of 0.2 across the districts (Table 3). This parameter differed substantially among 13 districts studied. The Jiparana district of Rondonia showed the highest estimated heterozygosity over 62 polymorphic loci, whereas the Assis-Brasil district of Acre showed the lowest one. In spite of high standard errors of mean heterozygosities, probably due to the small sample sizes, recorded in Assis-Brasil (Acre), Ariquemes (Rondonia), and Vila Bela (Mato Grosso) populations, the remarkable variations of mean heterozygosity clearly showed differences in genetic variability among 13 districts. The average degree of diversity within individual districts using Shannon's diversity index (Shannon and Weaver, 1949) was 0.296 and ranged from 0.093 for the Assis-Brasil district of Acre to 0.389 for the Jiparana district of Rondonia (Table 3). Shannon index was correlated strongly with the percentage of polymorphic loci in a district. In fact, the districts with high Shannon's diversity index also exhibited the high percentages of polymorphic loci (Table 3). Similar to mean heterozygosity, differences in values of Shannon's diversity index also showed genetic differentiation among the districts. The largest average genetic distance among accessions within the districts was detected in the Jiparana district of Rondonia and the smallest was found in the Assis-Brasil district of Acre, of which the average genetic distance values were 0.469 and 0.139, respectively (Table 3). According to previous studies, high genetic diversity is usual in IRRDB'81 accessions (Chevallier, 1988; Besse et al., 1994; Lekawipat et al., 2003). All the accessions had unique RAPD genotypes. Nei's genetic distance values between pairs of districts ranged from 0.046 for Catriquacu and Itanba of Mato Grosso to 0.304 for Tarauaca of Acre and Aracatuba of Mato Grosso (Table 4). The dendrogram constructed by UPGMA cluster analysis showed that *Hevea* IRRDB'81 collection of 13 different districts were in five clusters with Ariquemes of Rondonia quite different from the others (Fig. 8). Among the districts of Mato Grosso, Vila Bela was in the same cluster with the districts from Rondonia (Jaru and Jiparana), meanwhile other districts (Itanba, Catriquacu and Aracatuba) were grouped into one cluster. This showed that Vila Bela was quite different from other districts of Mato Grosso, which was also noted by other researchers using RFLP markers (Besse et al., 1994). The distribution of those districts in genetic cluster analysis seemed to conform to geographical origins of *Hevea* IRRDB'81 collection, except Calama district of Rondonia.


Table 4. Nei's genetic distance among 13 districts of *Hevea* IRRDB'81 collection

Fig. 8. Dendrogram of UPGMA cluster analysis based on Nei's genetic distance

Analysis of molecular variance (AMOVA) for 59 accessions of 13 districts revealed that the interpopulation (interdistrict) variation accounted for only 14.1% of the total genetic variance while intrapopulation (within district) variation accounted for 85.9% (Table 6). The results indicated that the majority of genetic diversity was within district variation (85.9%). The results suggests that there might be a certain gene flow among the districts, possibly owing to the species' outcrossing as a mode of reproduction and dispersion of seeds by a network of rivers in Amazon basin. However, it should be treated with caution due to small samples brought into the study.


*Significant at P=0.001* 

446 Genetic Diversity in Plants

Among the districts of Mato Grosso, Vila Bela was in the same cluster with the districts from Rondonia (Jaru and Jiparana), meanwhile other districts (Itanba, Catriquacu and Aracatuba) were grouped into one cluster. This showed that Vila Bela was quite different from other districts of Mato Grosso, which was also noted by other researchers using RFLP markers (Besse et al., 1994). The distribution of those districts in genetic cluster analysis seemed to conform to geographical origins of *Hevea* IRRDB'81 collection, except Calama district of

**District AC/AB AC/B AC/F AC/S AC/T MT/A MT/C MT/IT MT/VB RO/A RO/C RO/J RO/JP AC/AB** 0.147 0.201 0.146 0.160 0.297 0.244 0.236 0.202 0.286 0.188 0.236 0.266 **AC/B** 0.147 0.106 0.107 0.155 0.190 0.135 0.161 0.216 0.275 0.135 0.160 0.164 **AC/F** 0.201 0.106 0.105 0.145 0.125 0.115 0.129 0.213 0.230 0.107 0.144 0.150 **AC/S** 0.146 0.107 0.105 0.190 0.153 0.151 0.146 0.158 0.239 0.103 0.164 0.169 **AC/T** 0.160 0.155 0.145 0.190 0.304 0.227 0.234 0.193 0.236 0.230 0.205 0.212 **MT/A** 0.297 0.190 0.125 0.153 0.304 0.094 0.086 0.296 0.295 0.171 0.191 0.183 **MT/C** 0.244 0.135 0.115 0.151 0.227 0.094 0.046 0.202 0.286 0.125 0.145 0.141 **MT/IT** 0.236 0.161 0.129 0.146 0.234 0.086 0.046 0.189 0.289 0.147 0.145 0.147 **MT/VB** 0.202 0.216 0.213 0.158 0.193 0.296 0.202 0.189 0.233 0.123 0.146 0.149 **RO/A** 0.286 0.275 0.230 0.239 0.236 0.295 0.286 0.289 0.233 0.222 0.189 0.205 **RO/C** 0.188 0.135 0.107 0.103 0.230 0.171 0.125 0.147 0.123 0.222 0.137 0.172 **RO/J** 0.236 0.160 0.144 0.164 0.205 0.191 0.145 0.145 0.146 0.189 0.137 0.097 **RO/JP** 0.266 0.164 0.150 0.169 0.212 0.183 0.141 0.147 0.149 0.205 0.172 0.097

Table 4. Nei's genetic distance among 13 districts of *Hevea* IRRDB'81 collection

Fig. 8. Dendrogram of UPGMA cluster analysis based on Nei's genetic distance

Rondonia.

Table 5. Summary of the AMOVA analysis

The IRRDB'81 *Hevea* germplasm in Vietnam exhibited large genetic variability by RAPD markers. The intradistrict source accounted for most of the genetic variation of the germplasm. Based on the genetic distance analysis, the collection could be classified into five groups which could help in planning crosses based on genetic distance in the hope of looking for heterosis and maintaining the abundant genetic diversity. The RAPD markers could also help in checking the genetic variability of the *Hevea* breeding program. Moreover, in association with the agronomical characteristics, morphological traits and isozymes analysis, RAPD markers are now suitable tools for genetic diversity studies of *Hevea* germplasm and can be useful for accumulation and management of genetic-breeding resources of *Hevea brasiliensis*.

#### **5. Utilization of** *Hevea* **germplasm in Vietnam**

One of the major objectives of conservation of *Hevea* genetic resources is to make genetic diversity available for immediate or future use. The widest possible range of the genetic diversity has to be conserved in order to meet future, as yet unknown, needs. *Hevea* germplasm conservation program in Vietnam is expected to promote and facilitate the use of conserved materials through the maintenance of healthy and readily accessible and adequately characterised/evaluated materials, and proper documentation of the relevant information. Evaluation data of the agronomic performances and the morphological characteristics gathered during cultivation of the accessionshave been being recorded continuously using a specifically constructed program (Fig. 9). Currently, a total of about 3,500 *Hevea* accessions have been collected and *ex situ* conserved. This germplasm comprises of three main genetic resources: the Amazonian (A) (most of which belong to IRRDB'81 collection collected in the Amazonian habitats of the genus), the Wickham (W), and the Wickham x Amazonian (WA) resources. The majority of this germplasm were derived from the IRRDB'81 collection with a total of 2,999 accessions, each of which is a clone derived from originally collected seedlings. Most of them have been evaluated for the agronomical and morphological characteristics. In the view of the limitations of the agronomical and morphological traits, isozyme and RAPD markers were used to analyze the genetic diversity and structure of the IRRDB'81 collection for more effective utilization of the germplasm in *Hevea* breeding programs in Vietnam.

Since the IRRDB'81 collection exhibited very poor profiles in agronomical characteristics, especially latex productivity, the chance for direct use of this collection for latex purposes seemed to be impractical although certain accessions could be planted for timber purposes. Regarding widening the genetic base for genetic improvement, several promising IRRDB'81 accessions have been included in hand pollination program in RRIV since 1997. Based on the agronomical and morphological traits as well as the genetic diversity analysis, recently, many attempts have been made to enlarge the genetic base of *Hevea* breeding materials by polycrossing among different genetic resources. In this way, many crosses between maternal

Fig. 9. Genebank documentation program in conservation of *Hevea* germplasm

W and paternal A accessions have been preferably made and the progenies of which are in various phases of evaluation with the expectation that the W x A progenies could combine the good agronomical characteristics from parents while optimizing the genetic variability in this population. In fact, Amazonian accessions were crossed with Wickham high yielding and good set fruit clones such as PB260 and RRIC110, and progenies derived from these hand pollination crosses were disbudded into fields of early selection trials for evaluation of agronomical performances such as latex yield, growth and diseases incidence. In general, these progenies exhibited rather good in girth but very poor in latex productivity. Most of the progenies produced a very little or negligible amount of latex; this agreed closely with the previous finding in Ivory Coast for the progenies from W x A crosses (Clement-Demange et al, 1990). However, some progenies had the production of 1.9 – 3.2 gram/tree/tapping or 106 – 132% of the production of the control clone (i.e. PB 260), and a large number of the progenies showed very good in growth performance with girth at 34 months old after planting ranging from 15.6 – 21.2 cm, or 101 – 161% of that of the control clone. These progenies are being further tested in the small scale clonal trials and would be included as parents in future breeding programs. This gives a way for the opportunity of genetic improvement in breeding programs, particularly in latex productivity of the *Hevea* IRRDB'81 collection. In diseases incidence, all of the progenies exhibited varying degrees of infection to powdery mildew. The progenies derived from different paternal accessions showed significant differences in susceptibility to powdery mildew. For instance, the progenies derived from AC6/23 and AC35/114 were more susceptible to powdery mildew than other progenies. In contrast, the progenies derived from RO44/268 and RO44/71 seemed to be lightly susceptible to powdery mildew in comparison to other progenies. This result has contributed to the development of clonal disease resistance by genetic recombination using IRRDB'81 collection as paternal clones in *Hevea* breeding programs.

The results obtained so far can be considered as a basis to continue combining Wickham and IRRDB'81 genetic resources in breeding programs at RRIV. The W x A progeny population provides a valuable source for selecting multi-clones recommended for developing rubber cultivation in non-traditional regions in Vietnam.
