Genetic diversity of forest arabica coffee (Coffea arabica L.) in Ethiopia as revealed by random amplified polymorphic DNA (RAPD) analysis

Transcription

1 Hereditas 138: (2003) Genetic diversity of forest arabica coffee (Coffea arabica L.) in Ethiopia as revealed by random amplified polymorphic DNA (RAPD) analysis ESAYAS AGA 1,2, TOMAS BRYNGELSSON 1, ENDASHAW BEKELE 2 and BJO RN SALOMON 1 1 Department of Crop Science, Swedish Uni ersity of Agricultural Sciences, Alnarp, Sweden 2 Department of Biology, Addis Ababa Uni ersity, Addis Ababa, Ethiopia Aga, E., Bryngelsson, T., Bekele, E. and Salomon, B Genetic diversity of forest arabica coffee (Coffea arabica L.) in Ethiopia as revealed by random amplified polymorphic DNA (RAPD) analysis. Hereditas 138: Lund, Sweden. ISSN Received July 4, Accepted February 7, 2003 Genetic diversity within the forest Coffea arabica L. gene pool in Ethiopia has not been extensively examined with molecular markers. In the present study, a total of 75 polymorphic RAPD bands generated by twelve random primers were used to assess genetic diversity among 144 genotypes representing 16 C. arabica populations. The number of polymorphic bands detected with each primer ranged from 2 to 9 with a mean of 6.25 bands per primer. Banding patterns ranged in percentage polymorphism from 37 % to 73 % with an overall mean of 56 % for the populations analyzed. The amount of genetic variation among populations estimated by Shannon-Weaver diversity index was (H=0.30). The within population and between populations differentiation values were 0.65 and 0.35, respectively. Genetic differentiations within and between zones of sample collection sites were 0.80 and 0.20, respectively. Within population average similarities estimated by simple matching coefficients ranged from 0.72 to 0.85, with an overall average of In the cluster analysis that used individual samples as operational taxonomic units, most of the representatives of the same population failed to cluster before they joined members of other populations. Nevertheless, most of the populations were clustered on the basis of their geographic closeness and an east west differentiation was observed at approximately 75 % similarity. The results obtained provide information on how to select sites for in situ conservation of C. arabica germplasm. Esayas Aga, Swedish Uni ersity of Agricultural Sciences, Department of Crop Science, Box 44, SE Alnarp, Sweden. esayas.aga@vv.slu.se The genus Coffea L. consists of approximately 100 species (taxa) so far identified (BRIDSON and VER- COURT 1988). Coffee is an important agricultural export commodity in more than 50 developing countries of Africa, Asia and Latin America (RANI et al. 2000; OROZCO-CASTILLO et al. 1994; DUBLIN et al. 1991) and the commercial coffee production relies on two species of coffee, (C. arabica L. and C. canephora Pierre ex. Froehn.), with Coffea arabica being considered as superior quality coffee, and contributing to over 70 percent of the world s coffee production (RANI et al. 2000; OROZCO-CASTILLO et al. 1994). Coffea arabica is predominantly self-pollinating (autogamous) and the only natural allotetraploid (2n=4x=44) species in the genus Coffea. The other coffee species are all diploid (2n=2x=22) and out crossing (allogamous) (MEYER 1965; CHARRIER and BERTHAUD 1985). C. arabica is a perennial woody shrub with a dimorphic growth characteristic which consists of vertical (orthotropic) and horizontal (plagiotropic) branches. C. arabica has its origin in the Southwest highlands of Ethiopia (HARLAN 1969), but THOMAS (1942) reported the existence of wild Coffea arabica L. on the Boma plateau in Sudan. Yet it is possible that it had been introduced in the distance past from the main mass of the Abyssinian (Ethiopian) highlands (THOMAS 1942). BERTHAUD and CHARRIER (1988) also reported the presence of C. arabica L. populations on Mount Imantong in Sudan and Mount Marsabit in Kenya. It is generally believed that C. arabica has its highest genetic diversity in Ethiopia. The coffee in Ethiopia is grown under four different systems (forest coffee, small holder coffee, semi-plantation coffee and plantation coffee). Forest coffee, which is sometimes referred to as wild coffee accounts for 60 percent of coffee production and it is self sown seedlings which have been transplanted to give an irregular, but dominant understorey in the forest (VAN DER GRAAFF 1981). The forest arabica coffee population in Ethiopia is important for the improvement of the crop. Natural populations are likely to be the source of new resistance genes needed to cope with future evolution in the pathogens of a crop. This will especially be the case if host populations are maintained in situ to allow co-evolution to occur. Because almost all forest arabica coffee populations are owned by individual farmers, who may replace them either by resistant cultivars or other cash crops creating management difficulties if all populations of forest coffee are

2 Hereditas 138 (2003) Genetic di ersity of Coffea arabica as re ealed by RAPD 37 aimed to be preserved. As a result, preserving all populations is not possible and decisions about which populations to preserve must be made. Therefore, information on the amount and pattern of genetic variation of forest coffee populations in Ethiopia is a crucial variable in the planning process. At present, our knowledge on the amount and pattern of molecular genetic variation in forest coffee is low. Traditionally the genetic diversity assessments of agricultural species are based on morphological and agronomical characteristics. Although there is substantial intra-specific variation in vegetative traits, especially leaf and fruit characters, it is difficult to distinguish genotypes on their external morphology alone (CASAS et al. 1999) as phenotypic characters are generally influenced by environmental factors and growth stages of the plant. Morphological characters in perennial plants like coffee often require a lengthy and expensive evaluation during the whole vegetative growth. A variety of molecular techniques have been developed to measure genetic variation at both interspecific and intraspecific levels in a number of plant species. The random amplified polymorphic DNA (RAPD) marker technique (WELSH and McCLELLAND 1990; WILLIAMS et al. 1990) is quick, easy and requires no prior sequence information. RAPDs have been extensively used in clarifying plant taxonomic relationships at the level of species, subspecies or varieties (DIAZ and AGUINAGALDE 1996; VAN BUREN et al. 1994). In coffee, the RAPD technique has been used to study the genetic diversity and relationships among Coffea species (LASHERMES et al. 1993, 1996a; OROZCO- CASTILLO et al. 1994; ANTHONY et al. 2001), and for identification of resistance to coffee berry diseases (AGWANDA et al. 1997). In this paper a RAPD based genetic diversity study of forest coffee populations from major coffee producing regions in Ethiopia is presented. The results may provide information to select sites of greater genetic diversity for in situ conservation. MATERIAL AND METHODS Plant material Sixteen populations of forest arabica coffee (Coffea arabica L.) collected from four zones of the Oromiya region of Ethiopia were included in this study (Fig. 1). Seeds from 20 individuals were sampled randomly across a transverse pass through a sample area of approximately 20 by 20 m (400 m 2 ) for each population during the period of November to January 1999 (Table 1). The population size in the sampling area ranges from approximately 200 to over 300 individu- Fig. 1. Map of Ethiopia showing administrative regions and zones from which forest arabica coffee (C. arabica L.) samples were collected. Numbers in circles designate zones of sample collection sites. 1 Western Welega, 2 Ilubabor, 3 Jima, and 4 Bale. als with an average distance of 1.5 meter between each sample. The seed samples were germinated and grown in pots in a glass house. The leaf samples were collected from three to four weeks old germinated seedlings and was used for DNA extraction. DNA extraction Total DNA was extracted from fresh young leaf tissue of coffee plants following the CTAB procedure (WANG et al. 1996) with some modifications. Young leaf samples were frozen in liquid nitrogen and ground to powder. Powdered leaves ( g) were collected in eppendorf tubes and 750 l of extraction buffer (0.1 M Tris ph 7.5, 0.05 M EDTA, 0.5 M NaCl and 100 l 10%w/v SDS) was added. The mixture was incubated for 20 min at 65 C, 250 l of5mkac was added and kept on ice for 1 hour before centrifugation at rpm. The supernatant was precipitated with an equal volume of iso-propanol and centrifuged. The pellet was air dried, dissolved in 250 l of TE (10 mm Tris HCl, ph 7.6, 1 mm EDTA), 250 l of CTAB buffer (0.2 M Tris ph 7.5, 50 mm EDTA, 2 M NaCl, 2 % CTAB), and incubated for 15 min at 65 C. DNA was extracted twice with an equal volume of chloroform. The final water-phase was precipitated with iso-propanol and the DNA pellet was washed twice with 70 % ethanol. The DNA pellet was airdried, dissolved in 100 l ofte,5 l RNase (1 mg/ml) added, incubated at 37 C for 30 min, and kept at 20 C for later use. The DNA quality was checked by electrophoresis in a 1 % agarose gel and the concentration was estimated in relation to the concentration of co-migrating -phage DNA and by repeated measurements with spectrophotometer at 260 nm.

3 38 E. Agaetal. Hereditas 138 (2003) Table 1. List of forest arabica coffee (Coffea arabica L.) populations considered in the present study Population Sample Administrative Specific localities descriptions size zone of collection site Welega-1 9 West Welega Anfilo, Phali forest, 3 km north of Shola village Welega-2 9 West Welega Sidi forest, 10 km North of Sidi village Welega-3 9 West Welega Kerache forest, around Ega river 20 km south of Mugi town Walega-4 9 West Welega Around Sako river 20 km west of Dambi Dolo town Ilubabor-1 9 Ilubabor Mekanisa forest, 10 km west of Bedele town Ilubabor-2 9 Ilubabor On the descending slope to Gaba river bridge on the way from Bedele town Ilubabor-3 9 Ilubabor Dawe area, 20 km away from Gaba river bridge on the way to Matu town Ilubabor-4 9 Ilubabor On the ascending slope to Gore town on the way from Matu town Jima-1 9 Jima Beleta forest, 48 km south of Jima town on the way to Bonga town Jima-2 9 Jima Akato forest, 30 km west of Jima town on the way to Agaro Jima-3 9 Jima Near Gojab river bridge on the way to Bonga town from Jima town Jima-4 9 Jima Sema forest, 57 km west of Jima town on the way to Bedele town Bale-1 9 Bale Harena forest, 14 km north of Mena town on the way to Goba town Bale-2 9 Bale Harena forest, 37 km north of Mena town on the way to Goba town Bale-3 9 Bale Harena forest, 25 km west of Mena town Bale 4 9 Bale Harena forest, 20 km north east of Mena town Screening of RAPD primers A total of 144 individual plants representing 16 populations (nine individuals/population) were assayed. An initial screening of 100 oligonucleotide primers (A, B, C, F and G kits, and additional twelve individual primers (LASHERMES et al. 1996a), all from Operon Technologies (CA, USA) was carried out on two individuals from each population to identify primers that detect polymorphisms within or between populations. Out of the 112 primers tested, only twelve primers produced reproducible variation and were used in further analyses (Table 2). The reproducibility of the banding patterns of each primer was tested with respect to the amplification conditions, concentration of the primer relative to the template DNA and magnesium chloride concentration. Once optimal conditions had been determined for each specific primer, the conditions were strictly followed. PCR amplification The DNA amplification reactions were performed in a total volume of 20 l containing 1 reaction buffer (75 mm Tris-HCl, ph 8.8, 20 mm (NH 4 ) 2 SO 4, 0.01 % (v/v) Tween 20), 2.25 mm MgCl 2, 12 ng primer, 0.4 mm dntps (100 M each of datp, dctp, dgtp and dttp), 0.6 units of Taq polymerase, and ng of sample DNA. A master mix was prepared for each primer to minimize measurement deviation. The reaction mixtures were overlaid with two drops of light mineral oil. Amplification was carried out in a Hybaid Omnigene thermocycler with one cycle of initial strand separation at 94 C for 3 min followed by 45 cycles of 1 min at 94 C, 1 min at 37 C, and 2 min at 72 C using the fastest available temperature transitions. The last cycle was followed by an additional extension at 72 C for 10 min to ensure that the primer extension reaction was completed. The PCR amplification products were separated on 1.4 % agarose gels containing 0.5 g/ml ethidium bromide, and run in 1 TAE buffer (40 mm Tris acetate ph 8.0, 1 mm EDTA) at 90 volts. A hundred base pair DNA ladder was used as a molecular weight marker. The reproducibility of the amplification products was checked two times for each primer. A control containing all components of a typical reaction but lacking template DNA was used. Those fragments that were monomorphic, not reproducible, appear in the control reaction or too difficult to score with certainty were excluded from the data analysis. Data scoring and analysis Each amplified DNA fragment was considered as an independent character (locus), and scored as present (1) or absent (0). Each amplified product was named by the code of the primer followed by its size in base pairs. Since RAPD markers are dominant, a locus was considered to be polymorphic if the presence and absence of the bands were observed in various individuals, and monomorphic if the bands were present in all individuals. No distinction was made between fragments of the same molecular size that varied in intensity. The magnitude of genetic variation was determined for each population and each zone using the Shannon s diversity index described by HUTCHENSON (1970), which is given as: H= P i ln P i, where P i is the proportion of amplified bands among individuals of a population. Shannon s

4 Hereditas 138 (2003) Genetic di ersity of Coffea arabica as re ealed by RAPD 39 Table 2. List of selected primers used in the RAPD analysis along with their nucleotide sequences, number of polymorphic bands and estimated molecular size range Primer Sequences 5 to 3 Number of polymorphic bands Molecular size range OPA07 GAAACGGGTG OPA15 TTCCGAACCC OPB02 TGATCCCTGG OPB13 TTCCCCCGCT OPC07 GTCCCGACGA OPC10 TGTCTGGGTG OPC15 GACGGATCAG OPF05 CCGAATTCCC OPI20 AAAGTGCGGG OPN18 GGTGAGGTCA OPN20 GGTGCTCCGT OPX20 CCCAGCTAGA Total 75 Range Average 6.25 diversity index is frequently used in RAPD data analysis because the index is insensitive to bias that may be introduced into data due to undetectable heterozygosity (GUSTAFSON et al. 1999; MAKI and HORIE 1999; OIKI et al. 2001). Following the method of KING and SCHAAL (1989), we estimated the partitioning of genetic variation into within and between populations from Shannon s diversity data. The proportions of variation attributed to differentiation within populations and within zones as well as differentiation between populations and between zones were estimated from the Shannon-Weaver diversity index. Simple matching coefficient (SOKAL and MICH- ENER 1958) was used to compute similarity between each pair of individuals. Similarity between each pair of populations was also computed from the raw data using the Dice similarity coefficient (DICE 1945), which is equivalent to NEI and LI (1979) as both calculate similarities from shared presence and not absence of DNA bands. Inter-population similarity was calculated as the mean of S ij of all different comparisons between plants from each pair of populations. The similarity matrices computed for each pair of individuals as well as each pair of populations were subjected to cluster analysis using the unweighted pair-group method with arithmetic averages (UPGMA) (SNEATH and SOKAL 1973), and dendrograms were generated using NTSYS-PC version 1.8 (ROHLF 1993) to visualize the genetic similarity between individuals, and between populations. In addition, since the cluster analysis of all individuals is very complex, we chose three individuals at random from each population to visualize individual relationships in a dendrogram. This was compared to the dendrogram constructed at the population level. Cophenetic values were computed for each tree matrix and compared to the actual similarity matrix to evaluate the degree of fitness between the two matrices. RESULTS Out of the 112 primers initially surveyed, RAPD patterns from twelve primers were found to be reproducible and suitable for this investigation. The twelve oligonucleotide primers generated a total of 75 stable (reproducible) polymorphic bands across 144 individuals representing 16 populations. The number of bands per primer varied from two (OPX20) to nine (OPC07) with an average of 6.25 per primer, and the estimated molecular size was in the range of 300 to 2400 base pairs (Table 2). Fifty two percent of the polymorphic bands have a frequency greater than 0.50 with an overall frequency range of 0.03 to Fig. 2 shows a representative picture of the electrophoretic pattern of PCR amplified DNA fragments obtained during the analysis using the OPC15 primer. The proportion of polymorphic bands within populations varied from 37 % for Jima-3 to 73 % for Jima-2 populations, with a mean of 56 % (Table 3). The estimated Shannon s diversity index (H) for each population computed across primers varied from for Welega-1, Jima-1 and Jima-3 populations to for Welega-2 population with an overall mean across populations (H pop ) of The mean for the entire data, when individuals

5 40 E. Agaetal. Hereditas 138 (2003) Table 4. Partitioning of the genetic ariation into within and between populations as well as within and between zones of sample collection sites Category Parameter Mean S.E S.D Fig. 2. Example of electrophoretic pattern of PCR amplified DNA fragments of C. arabica populations produced by RAPD primer OPC15. Lane 1 is a molecular size marker and lane 20 is a negative control. Lanes 2 5, lanes 6 8, lanes 9 13, and lanes are Welega, Ilubabor, Jima, and Bale populations, respectively. of all populations were considered together (H sp )was (Table 4). For the zones of sample collection sites the values range from to for Bale and Welega zones, respectively (Table 3). The differentiation of genetic variation within and between populations revealed that the within population variation accounted for 65 percent and the remaining 35 percent occurred between populations. Population Zone H pop H sp H pop /H sp (H sp H pop )/H sp H zone H zone /H sp (H sp H zone )/H sp H pop and H zone =Mean genetic variation for the populations and zones, respectively. Hsp=Mean for genetic variation computed from the entire data, when individuals of all populations were considered together. H pop /H sp and H zone /H sp =Proportion of genetic variation within populations and zones, respectively. (H sp H pop )/H sp and (H sp H zone )/H sp =Proportion of genetic variation between populations and zones, respectively. On the other hand, the genetic differentiation within and between zones of sample collection sites was 0.80 and 0.20, respectively. Simple matching similarity coefficients between all possible pairs of genotypes varied from 0.48 between Table 3. Proportion of polymorphic bands, mean similarity coefficient within populations, and mean estimates of the Shannon-Wea er di ersity index for the 16 forest arabica coffee populations and their four zones of collection sites Categories Polymorphic Mean similarity Shannon-Weaver diversity index bands (%) Within populations (Mean S.E) S.D Population Welega Welega Welega Welega Ilubabor Ilubabor Ilubabor Ilubabor Jima Jima Jima Jima Bale Bale Bale Bale Mean Entire data Zone Welega Ilubabor Jima Bale Mean

6 Hereditas 138 (2003) Genetic di ersity of Coffea arabica as re ealed by RAPD 41 genotypes of Ilubabor-2 and Jima-4 populations to 0.97 between genotypes of Bale-3 population. The mean within similarity indices presented by the 16 coffee populations ranged from 0.72 to 0.85, with an overall mean of 0.78 (Table 3), which means that the individuals from each coffee population share, on average, 78 percent of their RAPD fragments. The degree of within variability presented by some populations such as Welega-2, Jima-2 and Jima-4 populations was so elevated that they were higher than some between population differentiation values. The between population similarities calculated by the Dice similarity coefficient indicated that the mean similarity among all populations was 0.72, and the mean similarity of each population with all others ranged from 0.68 to Considering the pair-wise comparisons of all populations, the range was not very extensive with the lowest value of 0.61 and the highest of 0.86 (Table 5). Cluster analysis The result of the cluster analysis obtained using all individuals is similar to the clustering of the randomly chosen 48 individuals (three from each population), in which two tied trees and two major clusters were observed (Fig. 3). Cluster one consists of individuals from the Welega and Ilubabor populations, which formed sub-clusters partially on the basis of their geographical proximities. In the second cluster, most of the individuals from the Jima and Bale populations were clustered on the basis of their respective zones of sample collection sites. In spite of the low degree of variability detected in all populations, the phenetic analysis made using individual plants as operational taxonomic units demonstrated that most of the individuals fail to cluster on the basis of their respective populations. The cluster analysis at the population level based on Dice coefficients of similarity also displayed a similar pattern in which two main clusters were revealed (Fig. 4). Cluster (1) includes the Welega and Ilubabor populations, which were further grouped into two sub-clusters and an outlier (Welega-2 population) at 74 % similarity. Cluster (2) consists of the Jima and Bale populations, which were further differentiated on the basis of their zone of sample collection sites at 75 % similarity. Although the pair wise similarity coefficient between individuals ranged from 48 % to 97 %, the dendrogram displayed that there is no observed differentiation between any two individuals at a similarity below 61 %. The correlation coefficient ( r value) for the clustering at individuals and populations level was 0.80 indicating a good fit of the cluster analysis performed. Table 5. Dice coefficient of similarity between the 16 forest arabica coffee populations Welega-1 Welega Welega Welega Ilubabor Ilubabor Ilubabor Ilubabor Jima Jima Jima Jima Bale Bale Bale Bale Mean Overall average 0.72

7 42 E. Agaetal. Hereditas 138 (2003) Fig. 3. Dendrogram generated using UPGMA demonstrating the genetic similarities between three individuals randomly selected from each of the 16 forest arabica coffee (C. arabica L.) populations. Abbreviations refer to the first three letters and number with which a population is designated (Table 1). DISCUSSION Artifactual variation represents a potential problem in surveys of genetic variation in natural populations and must be discriminated from true polymorphism for the applications of RAPD to be both accurate and reliable (ELLSWORTH et al. 1993). In the present study, the effect of artifacts produced was minimized using spectrophotometry to standardize the DNA concentrations across all individuals. In addition, the DNA samples were RNase treated to avoid RNA to influence determination of DNA concentration.

8 Hereditas 138 (2003) Genetic di ersity of Coffea arabica as re ealed by RAPD 43 Fig. 4. Dendrogram generated by UPGMA cluster analysis based on genetic diversity of the 16 forest arabica coffee (C. arabica L.) populations. The data obtained in the present study shows low to moderate levels of polymorphism within, and among forest C. arabica populations in Ethiopia. Extent of distribution, areas sampled and plant characteristics such as mode of reproduction, breeding behavior and generation time are some of the important parameters that determine the level of genetic variability revealed in a species (BHAT et al. 1999). Because C. arabica is an allotetraploid and a predominantly self-pollinated species a high degree of genetic uniformity is expected (LASHERMES et al. 1996a). The RAPD markers showed that the Jima-2 population had the highest percent of polymorphic bands (73 %) while Jima-3 had the lowest percent (37 %) with an overall mean percent polymorphic bands of 56 %. On the other hand, in the diversity indices of Shannon-Weaver (H), the Welega-2, Welega-4 and Jima-2 populations, showed a relatively higher level of variability with diversity indices of 0.41, 0.39 and 0.38, respectively. The lowest variability was observed in Welega-1, Jima-3 and Jima-4 populations with diversity indices of each being On regional (zone of sample collection sites) basis, the Welega and Ilubabor zones had higher values (0.42 and 0.40, respectively) while the Bale zone had the lowest value (0.31) of the Shannon-Weaver diversity indices. The differentiation of variation into within population and between populations showed that most of the variation was distributed within populations (65 %) rather than between populations (35 %). Since one expects a greater value of between populations differentiation in predominantly self-pollinating or geographically isolated species (OIKI et al. 2001), the inter-population genetic differentiation observed in the present study is low compared to 46.1 % reported for the self-pollinated species Campanula microdonta (OIKI et al. 2001). The present study demonstrates higher values of genetic differentiation within zones of sample collection sites than between zones. AYANA et al. (2000a,b) reported a similar pattern of differentiation for both cultivated sorghum (Sorghum bicolor (L.) Moench) and wild sorghum (Sorghum bicolor ssp. erticilliforum (L.) Moench) In order to quantify the level of polymorphism between forest C. arabica populations, the Dice coefficient of similarity was used to generate a similarity matrix (Table 5). One single tree was identified by the UPGMA method (Fig. 4). The correlation between the cophenetic value matrix and the actual matrix was moderately high (0.80) indicating a good fit of the cluster analysis performed. The absolute genetic distance between all possible pairs of 144 individual plants estimated as 1 S ij from SMC was in the range of 0.03 to The genetic distance at the population level fell in the range of 0.14 to 0.39, indicating low to moderate polymorphism. LASHER- MES et al. (1996a) reported a comparable level of molecular polymorphism in wild accessions of C. arabica. ANTHONY et al. (2001) also reported a comparable level of polymorphism among 80 accessions of C. arabica L. derived from spontaneous and subspontaneous trees in Ethiopia. However, in the present report, out of the primers previously used by LASHERMES et al. (1996b), only four primers (OPI20, OPN18, OPN20 and OPX20) showed variation among the forest C. arabica samples. Considering the result of these four primers alone, the mean percent polymorphism was 64 % for the populations analyzed, and the mean Shannon-Weaver diversity indices was H =0.29. The low level of polymorphism detected in

9 44 E. Agaetal. Hereditas 138 (2003) C. arabica was related to its autogamous and allotetraploid origin (LASHERMES et al. 1996a). Low level of DNA polymorpism was also reported in several other self-pollinating crops such as wheat (JOSHI and NGUYEN 1993), pigeonpea (RATNAPARKHE et al. 1995), and tomato (WILLIAMS and CLAIR 1993). However, the polymorphism detected in the present forest C. arabica samples is not comparable to the level of polymorphism detected in the above crops, because the present study is not based on cultivar identification or on morphological character variation, rather it is based on random samples of forest coffee collected from various regions. On a regional base, the Bale (Harana forest) zone presented the lowest values of Shannon-Weaver diversity indices and relatively higher values of within zone similarity coefficient. We suggest that the genetic uniformity found in this area could be due to a very narrow genetic base originating from a few plants growing in the area. Nevertheless, we witness that coffee samples from the Harana forest is a part of the primary forest ecosystem, and could be the true natural forms of forest (wild) C. arabica. The presence of a band at 700 bp with the primer OPC15 (Fig. 2) and absence of bands at 1100 bp and 350 bp with the primers OPN20 and OPX20 (data not shown), respectively, were specific in most of the forest C. arabica samples from the Harana forest. Only one sample from the Jima-1 population had the OPC bp band (Fig. 2). This result agrees with differentiation of C. arabica germplasm accessions into east and west of Great Rift Valley, previously established by analysis with RAPD markers (LASH- ERMES et al. 1996a) and by a multivariate analysis of phenotypic characters (MONTAGNON and BOUHAR- MONT 1996). Since phylogenetic studies based on cpdna suggested recent origin of the genus Coffea (LASHERMES et al. 1996b; CROS et al. 1998), the presence of the tectonic fault, the Great Rift Valley may not account for the distinction between the Southwestern and Southeastern forest C. arabica populations. However, in the present study, sample populations collected from Jima (West of Rift Valley) were found to cluster with sample populations collected from Bale (East of Rift Valley) (Fig. 4). Such clustering may not be explained in terms of proximity of origin, as the two zones of collection sites (Jima and Bale) are air distance estimates of more than 300 km apart. On the other hand, the low differentiation observed between Southwest and Southeast forest C. arabica populations may support the hypothesis that Southeastern coffee trees were introduced from the Southwest part of the country (ANTHONY et al. 2001). However, this may not rule out the possibility that C. arabica in Southeast could have been selected from the wild coffee growing locally, because primary forests with coffee trees still exist in the Southeastern region (e.g. Harana forest). It is obvious that all variation of a given species cannot be conserved. The conservation strategies will vary with the species under study and the primary interest (BROWN and BRIGGS 1991). In coffee, there have been no alternatives to ex situ field collections for long-term germplasm conservation, because coffee seeds are recalcitrant. If conventional methods of seed storage are used, the C. arabica seeds will be viable for a maximum of two to three years (VAN DER VOSSEN 1985). In ex situ field collections there is a risk of losing valuable germplasm due to diseases and pests as well as to poor adaptation of certain species to the local environment. DULLOO et al. (1998) have described strategies for in situ conservation of Coffea species. The in situ approach of plant genetic conservation has been recently emphasized (MAXTED et al. 1997). Ex situ and in situ conservation strategies are complementary and should not be viewed as antagonistic (NEVO 1998). The in situ method allows continuing evolution of the species in its natural habitat in order to allow perpetuation and integration of co-adapted gene complexes (MAXTED et al. 1997). The ex situ approach, on the other hand, safeguards the species genetic diversity in case of possible habitat destruction and represent a readily available source of germplasm for research and breeding. Since there is a considerable amount of ex situ germplasm of C. arabica collected and maintained in Ethiopia (BELLACHEW 1997), no further collection is needed except for peripheral areas such as the Harana forest, which is under various environmental threats. On the other hand, the selection of forest C. arabica populations for in situ conservation appears to be more important. The result of the present study has demonstrated that the RAPD technique could be applied for measuring the degree of variability within, and between forest C. arabica L. populations. Although the sample size analyzed was small for each population, we suggest that some sample populations have genetic variability that need to be conserved, of which, Walega-2, Walega-4 and Jima-2 populations should be given higher priorities. The Harana forest populations seem to be in the primary gene pool and may need particular attention with regard to either ex situ or in situ conservations. However, before making the final decision to select the forest C. arabica populations to be conserved in situ, we recommend further analysis using co-dominant molecular markers with a better resolution power, because RAPD markers are dominant markers in which the homozygous presence of a fragment is not distinguishable from its het-

10 Hereditas 138 (2003) Genetic di ersity of Coffea arabica as re ealed by RAPD 45 erozygote and as a result, it underestimates the level of existing variability. ACKNOWLEDGEMENTS This work was supported by BIO-EARN (East African Regional Programme and Research Network for Biotechnology, Biosafety and Biotechnology Policy Development) funded by SIDA/SAREC, Sweden. We acknowledge Mrs Britt Gre en, Dr. Amsalu Ayana, Dr. Karin Persson and Mr. Göran Olsson at the Department of Crop Science, Alnarp, Sweden, for technical assistance and taking care of our coffee plants. REFERENCES Agwanda CO, Lashermes P, Trouslot P, Comes MC and Charrier A, (1997). Identification of RAPD markers for resistance to coffee berry disease Colletotrichum kahawae, in arabica coffee. Euphytica 97: Anthony F, Bertrand B, Quiros O, Wilches A, Lashermes P, Berthaud J and Charrier A, (2001). Genetic diversity of wild coffee (Coffea arabica L.) using molecular markers. Euphytica 118: Ayana AA, Bryngelsson T and Bekele E, (2000a). Genetic variation of Ethiopian and Eritrean sorghum (Sorghum bicolour (L.) Moench) germplasm assessed by random amplified polymorphic DNA (RAPD). Genetic Resour. Crop Evol. 47: Ayana A, Bekele E and Bryngelsson T, (2000b). Genetic variation in wild sorghum (Sorghum bicolour ssp. Verticilliflorum (L.) Moench) germplasm from Ethiopia assessed by random amplified polymorphic DNA (RAPD). Hereditas 132: Bellachew B, (1997). Arabica coffee breeding in Ethiopia: a review. In: Proc. 17th ASIC Colloquium (Nairobi), p ASIC Paris France. Berthaud J and Charrier A, (1988). Genetic resources of Coffea. In: (eds RJ Clarke and R Macrae), Coffee, Vol. 4, Agronomy. Elsevier Applied Science, p Bhat KV, Babrekar PP and Lakhanpaul S, (1999). Study of genetic diversity in Indian and exotic sesame (Sesamum indicum L.) germplasm using random amplified polymorphic DNA (RAPD) markers. Euphytica 110: Bridson DM and Vercourt B, (1988). Rubiaceae (Part 2). In: Flora of tropical East Africa (ed. RM Polhill). Balkema, Rotterdam/Brookfield, p Brown AHD and Briggs JD, (1991). Sampling strategies for genetic variation in ex situ collections of endangered plant species. In: Genetics and conservation of rare plants (eds DA Falk and KE Holsinger). Oxford University Press, p Casas AM, Igartua E, Balaguer G and Moreno MA, (1999). Genetic diversity of Prunus rootstocks analysed by RAPD markers. Euphytica 110: Charrier A and Berthaud J, (1985). Botanical classification of coffee. In: Coffee, botany, biochemistry and production of beans and beverage (eds MN Clifford and KC Willson). Croom Helm, London and Sydney, p Cros J, Combes MC, Trouslot P, Anthony F, Hamon S, Charrier A and Lashermes P, (1998). Phylogenetic relationships of Coffea species: new evidence based on the chloroplast DNA variation analysis. Mol. Phylogenet. Evol. 9: Diaz Z and Aguinagalde I, (1996). The use of random amplified polymorphic DNA (RAPD) markers for the study of taxonomic relationships among species of Asphodelus sec. Verinea (Asphodelaceae). Am. J. Bot. 83: Dice LR, (1945). Measures of the amount of ecologic association between species. Ecology 26: Dublin P, Enjalric F, Lardet F, Carron MP, Trolinder N and Pannetier C, (1991). Estate crops. 1. Coffea species. In: Micropropagation (eds PC Debergh and RH Zimmerman). Kluwer Academic Publ, p Dulloo ME, Guarino L, Engelmann F, Maxted N, Newbury JH, Attere F and Ford-Lloyd BV, (1998). A complementary conservation strategies for the genus Coffea: A case study of Mascarene Coffea species. Genetic Resour. Crop Evol. 45: Ellsworth DL, Rittenhouse KD and Honeycutt RL, (1993). Artifactual variation in randomly amplified polymorphic DNA banding patterns. Biotech. 14: Gustafson DJ, Gibson DJ and Nickrent DL, (1999). Random amplified polymorphic DNA variation among remnant big bluestem (Andropogon gerardii Vitman) populations from Arkansas Grand Prairie. Mol. Ecol. 8: Harlan JR, (1969). Ethiopia: a center of diversity. Econ. Bot. 23: Hutchenson K, (1970). A test for comparing diversities based on the Shannon formula. J. Theoret. Biol. 29: Joshi CP and Nguyen HT, (1993). RAPD (random amplified polymorphic DNA) analysis based on intervarietal genetic relationships among hexaploid wheats. Plant Sci. 93: King LM and Schaal BA, (1989). Ribosomal-DNA variation and distribution in Rudbeckia missouriensis. Evolution 43: Lashermes P, Cros J, Marmey P and Charrier A, (1993). Use of random amplified DNA markers to analyse genetic variability and relationships of Coffea species. Genetic Resour. Crop Evol. 40: Lashermes P, Trouslot P, Anthony F, Comes MC and Charrier A, (1996a). Genetic diversity for RAPD markers between cultivated and wild accessions of Coffea arabica. Euphytica 87: Lashermes P, Cros J, Combes MC, Trouslot P, Anthony F, Hamon S and Charrier A, (1996b). Inheritance and restriction fragment length polymorphism of chloroplast DNA in the genus Coffea L. Theor. Appl. Genet. 93: Maki M and Horie S, (1999). Random amplified polymorphic DNA (RAPD) markers reveal less genetic variation in the endangered plant Cerastium fischerianum var. molle than in the widespread conspecific C.fischerianum (Caryophyllaceae). Mol. Ecol. 8: Maxted N, Ford-Lloyd BV and Hawkes JC, (1997). Plant genetic conservation. The in situ approach. Chapman and Hall. Meyer GF, (1965). Notes on wild Coffea arabica from Southwestern Ethiopia, with some historical considerations. Econ. Bot. 19: Montagnon C and Bouharmont P, (1996). Multivariate analysis of phenotypic diversity of Coffea arabica. Genetic Resour. Crop Evol. 43:

11 46 E. Agaetal. Hereditas 138 (2003) Nevo E, (1998). Genetic diversity in wild cereals: regional and local studies and their bearing on conservation ex situ and in situ. Genetic Resour. Crop Evol. 45: Nei M and Li WH, (1979). Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl Acad. Sci. USA 76: Oiki S, Kawahara T, Inoue K, Ohara M and Maki M, (2001). Random amplified polymorphic DNA (RAPD) variation among populations of the insular endemic plant Campanula microdonta (Campanulaceae). Ann. Bot. 87: Orozco-Castillo C, Chalmers KJ, Waugh R and Powell W, (1994). Detection of genetic diversity and selective gene introgration in coffee using RAPD markers. Theor. Appl. Genet. 87: Rani V, Singh KP, Shiran B, Nandy S, Goel S, Devarumath RM, Sreenath HL and Raina SN, (2000). Evidence for new nuclear and mitochondrial genome organizations among high-frequency somatic embryogenesisderived plants of allotetraploid Coffea arabica L. (Rubiaceae). Plant Cell Rep. 19: Ratnaparkhe MB, Gupta VS, Ven Murthy MR and Ranjekar PK, (1995). Genetic fingerprinting of pigeonpea (Cajanus cajan L. Millsp.) and its wild relatives using RAPD markers. Theor. Appl. Genet. 91: Rohlf FJ, (1993). NTSYS-pc. Numerical taxonomy and multivariate analysis system, version Applied Biostatistics Inc., New York. Sneath PHA and Sokal RR, (1973). Numerical taxonomy. The principles and practice of numerical classification. W. H. Freeman, p Sokal RR and Michener CD, (1958). A statistical method for evaluating systematic relationships. Univ. Kansas Sci. Bull. 38: Thomas AS, (1942). The wild arabica coffee on the Boma plateau, Anglo-Egyptian Sudan. The Empire J. Exp. Agri. 10: Van Buren R, Harper KT, Anderson WR, Stanton DJ, Seyoum S and England JL, (1994). Evaluating the relationship of Autumn buttercup (Ranunculus acriformis var. aestivalis) to some close congeners using random amplified polymorphic DNA. Am. J. Bot. 81: Van der Graaff NA, (1981). Selection of Arabica coffee types resistant to coffee berry disease in Ethiopia. PhD thesis, University of Wageningen, Netherlands, p Van der Vossen HAM, (1985). Coffee breeding and selection. In: Coffee, botany, biochemistry and production of beans and beverage (eds MN Clifford and KC Willson). Croom Helm, London and Sydney, p Wang M, Oppedijk BJ, Lu X, Duijn BV and Schilperoort RA, (1996). Apoptosis in barley aleurone during germination and its inhibition by abscisic acid. Plant Mol. Biol. 32: Welsh J and McClelland M, (1990). Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acids Res. 18: Williams CE and Clair DAS, (1993). Phenetic relationships and levels of variability detected by restriction fragment length polymorphism and random amplified polymorphic DNA analysis of cultivated and wild accessions of Lycopersicon esculentum. Genome 36: Williams JG, Kubelik AR, Livak KJ, Rafalski JA and Tingey SV, (1990). DNA polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Res. 18: