Somaclonal variation in Coffea arabica: effects of genotype and embryogenic cell suspension age on frequency and phenotype of variants

Tree Physiology 23, 419 426 2003 Heron Publishing Victoria, Canada Somaclonal variation in Coffea arabica: effects of genotype and embryogenic cell suspension age on frequency and phenotype of variants H. ETIENNE 1 3 and B. BERTRAND 2,4 1 Laboratorio de Biotecnología, CATIE, Apartado 11, 7170 Turrialba, Costa Rica 2 Present address: Centre de Coopération Internationale en Recherche Agronomique pour le Développement-Cultures Pérennes (CIRAD-CP), Genetrop, IRD, 911, Avenue Agropolis, BP 5045, 34032 Montpellier, France 3 Author to whom correspondence should be addressed (herve.etienne@cirad.fr) 4 Present address: Instituto Internacional de Cooperacion Agricola (PROMECAFE/IICA), Apartado 55, 2200 San José, Costa Rica Received April 5, 2002; accepted October 12, 2002; published online March 17, 2003 Summary We determined how age of embryogenic cell suspensions affects somaclonal variation in five F 1 hybrids of Coffea arabica L. Batches of plants were produced either directly from embryogenic callus, or after 3, 6, 9 and 12 months of embryogenic cell suspension culture. Seven phenotypic variants were characterized. Based on vigor and productivity of the regenerated plants, we classified the variants in order of increasing severity of physiological disorders as: Juvenile leaf color, Giant, Dwarf, Thick leaf (Bullata), Variegata, Angustifolia, and Multi-stem. The Dwarf, Angustifolia and Multi-stem variants were the most frequent among the regenerated plants (1.4, 4.8 and 2.9%, respectively). The frequency ( f ) of variants increased exponentially with the age (t) of the embryogenic suspension, in accordance with the function f = 0.99e 0.267t.For all genotypes, somaclonal variation was low (1.3%) in plants produced from embryogenic callus or 3-month-old cell suspensions and increased in frequency with increasing suspension age (6, 10 and 25% in plants produced from cell suspensions aged 6, 9 and 12 months, respectively). Large differences in somaclonal variation among genotypes were found only in plants produced from 12-month-old cell suspensions. For two genotypes, the oldest suspensions produced a majority of somaclonal variants (80 90%), whereas somaclonal variation ranged between 8 and 18% in the other genotypes. Cell suspension age and genotype also affected the type of variant produced. The severity of somaclonal variations increased with cell suspension age. For all genotypes combined, the Angustifolia variant was the most common. The other somaclonal variations were specific to certain genotypes or distributed randomly among the genotypes. Keywords: coffee, field trial, somatic embryogenesis. Introduction Coffea arabica L. varieties (self-fertilizing species, 2n =4x = 44) are traditionally propagated by seed. Given the strong hybrid vigor found when complementary genetic pools (Ethiopian Catimors) are crossed (Bertrand et al. 1999), substantial agronomic progress is expected from F 1 hybrids. Efficient micropropagation by somatic embryogenesis would facilitate large-scale dissemination of hybrid varieties in clonal form. There are several examples of this technique in small-scale projects. For instance, 19 clones of C. arabica F 1 hybrids were multiplied by somatic embryogenesis to set up assessment networks in Central America (Etienne et al. 2002). In Coffea canephora Pierre ex Froehn. var. robusta, a rootstock hybrid variety resistant to nematodes was made available on a commercial scale after multiplication of the two parents by somatic embryogenesis and the installation of large seed gardens in several Central American countries (Etienne et al. 2002). However, before the technique can be applied on an industrial scale, the processes need to be optimized, particularly with respect to: (1) a reduction in production costs and (2) a guarantee that the propagated trees are true-to-type. The second target can be achieved by reducing the occurrence of somaclonal variations at different stages of the procedure, through modifications of culture conditions and early selective elimination. In the cultivated species C. canephora and C. arabica, mass somatic embryo production has been carried out successfully in liquid medium (Starisky and Van Hasselt 1980, Zamarripa et al. 1991, Van Boxtel and Berthouly 1996). Recently, a procedure using a temporary immersion bioreactor was developed for C. arabica hybrids enabling mass and virtually synchronous production of germinated somatic embryos and efficient plant regeneration in the nursery after direct sowing (Etienne et al. 1997, Etienne-Barry et al. 1999). This procedure significantly reduces handling time, in vitro culture duration and shelving requirements, thereby reducing production costs. The existence of somaclonal variations in trees propagated by somatic embryogenesis has been demonstrated in C. arabica. Söndahl and Lauritis (1992) estimated that about 10% of trees regenerated from embryogenic callus were vari-

420 ETIENNE AND BERTRAND ants based on phenotypic characteristics. Recently, Etienne and Bertrand (2001) determined a variant frequency of about 2% in plants derived from 6-month-old embryogenic suspensions of four genotypes, based on easily identifiable morphological characteristics. These authors also showed that if the variants were discarded, there was no difference in the main agronomic characteristics between trees produced from embryogenic suspensions and control trees produced from microcuttings. There is little information on the impact of culture conditions on the occurrence of somaclonal variations during somatic embryogenic processes in coffee trees. Various reviews (Karp 1991, 1992, Buiatti and Gimelli 1993, Duncan 1997) have concluded that the frequency of somaclonal variations is related to: (1) explant source, ploidy level and chromosome number; (2) hormonal factors, i.e., the concentrations and types of growth regulators; (3) genotypic factors and (4) culture age. It is well known that the critical factor for somaclonal variation is the involvement of a disorganized growth phase, such as the proliferation of an embryogenic cell suspension (Karp 1991). The objective of our study was to determine how cell suspension age and genotype affect somaclonal variation of C. arabica embryogenic cell suspensions. We assessed death rates and the frequency and nature of variants produced by five hybrids under nursery conditions and in the field. Materials and methods Plant material Five clones and a control, comprising seedlings of the pure line, T8667 Catimor variety, were studied. The clones were derived from five F 1 hybrids of C. arabica obtained from crosses between the Caturra, Catimor (T8667) and Sarchimor (T5296) cultivated varieties with wild accessions from Ethiopia and Sudan: Caturra Ethiopian N 531 (Clone H1); T8667 Rume Sudan (Tree 1) (Clone H2), T8667 Rume Sudan (Tree 2) (Clone H3), T5296 Rume Sudan (Tree 1) (Clone H4), T5296 Rume Sudan (Tree 2) (Clone H5). Wild accessions were selected from the field collection at the Centro de Agronomia Tropical de Investigacion e Ensenanza (CATIE) in Costa Rica. Clones H2 and H3 thus came from the same F 1 family, as did Clones H4 and H5. Each clone was propagated by embryogenic cell suspension and the plants were regenerated every 3 months, i.e., after 0 (directly from embryogenic callus), 3, 6, 9 and 12 months of proliferation. Trial design For field assessments, plants were planted in five blocks, each block corresponding to plants derived from a particular embryogenic suspension proliferation time. The first block comprised 617 plants produced directly from embryogenic callus (0 months). The second, third, fourth and fifth blocks contained 757, 390, 494 and 826 plants derived from cell suspensions aged 3, 6, 9 and 12 months, respectively (see Table 3). Within the blocks, there were 15 trees per planting row. Because it was not possible to regenerate identical numbers of plants per clone and per suspension age, each row comprised plants from one clone. For all blocks combined, the total number of trees per genotype varied between 350 and 903 (see Table 4). The field trial was set up between June 1998 and July 1999 at 650 m above sea level in Turrialba, Costa Rica. The spacing was 1.80 m between rows and 0.9 m between trees within a row (6200 trees ha 1 ). Banana plants between the coffee rows at a density of one for every four coffee trees provided partial shade. Fertilization with N, P, K, Mg, B (18:3:10:8:0.5) was carried out in May and August at a rate of 1000 kg ha 1 year 1. A further 250 kg ha 1 year 1 of N was applied in November. To prevent fruit and leaf diseases the foliage was treated twice a year with copper hydroxide at a rate of 1.5 kg ha 1 year 1. In vitro multiplication techniques We used the embryogenic cell suspension method previously described by Etienne et al. (1997) and the culture media published by Van Boxtel and Berthouly (1996). The method involves four stages. In Stage 1, immature leaf explants are cultured for 1 month on medium containing 2.26 µm 2,4-D (2,4-dichlorophenoxyacetic acid), 4.92 µm IBA (indole-3-butyric acid) and 9.84 µm ip (iso-pentenyladenine), and then transferred for 6 months to a medium containing 4.52 µm 2,4-D and 17.76 µm BAP (6-benzylaminopurine), in order to develop an embryogenic callus on the explants. In Stage 2, the embryogenic callus is placed in liquid medium with 4.52 µm 2,4-D and 4.65 µm kinetin, producing a cell suspension of embryogenic aggregates. Long-term maintenance of the embryogenic suspension culture is achieved by 1-month proliferation cycles, i.e., twelve cycles were completed for the 1-year-old suspensions. In Stage 3, the embryogenic aggregates are transferred for 2 months to a temporary immersion bioreactor (RITA, CIRAD, France), containing a regeneration medium supplemented with 17.76 µm BAP. In Stage 4, the regenerated embryos are cultured on a germination medium containing 1.33 µm BAP. Plantlets are obtained from the germination medium after two consecutive 2-month subcultures in the bioreactor. Acclimatization Plantlets bearing 1 to 2 pairs of leaves were selected from the bioreactor. The base of each stem was cut and placed for 12 h in a rooting solution containing 492 µm IBA, 134 µm NAA (1-naphthalene acetic acid) and 23 µm kinetin. Plantlets were transferred to a horticultural substrate comprising 60% soil, 20% sand and 20% decomposed coffee pulp and placed in an acclimatization tunnel (80 90% relative humidity, 200 400 lux) for 1 month. Afterwards they were moved to a partially shaded nursery (2000 4000 lux) for 7 to 8 months. Acclimatized plants 30 to 40 cm in height were planted in the field. Trueness-to-type of micropropagated trees in the field Identification of variants was based on morphological observations after 8 months in the nursery and 2 years after planting in the field. Plants regenerated in vitro that had morphological TREE PHYSIOLOGY VOLUME 23, 2003

SOMACLONAL VARIATIONS IN COFFEE CELL SUSPENSIONS 421 traits different from those of the initial clone when acclimatized in the nursery or planted in the field were referred to as somaclonal variants. These somaclonal variants were identified based on height, morphology, leaf shape, productivity, fruit shape, leaf density, stomatal density and guard cell chloroplast number (cf. Krug et al. 1939). Variant frequencies were calculated based on the number of surviving plants. The mortality data represent cumulative nursery and field losses during the first two years following planting. Statistical analysis In all statistical analyses, the variable analyzed was the proportion p = X/n of variant individuals, where X is the number of somaclonal variants and n is the number of plants. For each proportion, a 3σ confidence limit for the binomial distribution was calculated as p ±3 pq n, where q is the proportion of normal plants. When two or more populations were compared, the data were organized in contingency tables. The test of independence was applied, calculating the quantity χ obs 2. The independence hypothesis was rejected when χ obs 2 2 χ 1 α. Yates s continuity correction was applied to the 2 2 contingency tables. For comparison of variants with normal trees, quantitative data were compared with those obtained from 10 normal trees (control). Means of the data from variants were compared to the means of data from controls by analysis of variance. Results Effects of genotype and cell suspension age on plant mortality In the field, mortality of control plants from seedlings was as high or higher than that of the cloned plants from somatic embryos (Table 1). Clones H1 and H2 had mortality rates significantly lower (P = 0.0001 and P = 0.06, respectively) than the controls. Among plants derived from somatic embryogenesis, the highest mortality was in plants derived from cell suspensions aged 0 and 9 months. In both cases, we attributed the high mortality to excess soil water due to heavy rainfall after planting. The lowest mortality was in plants derived from 12-month-old cell suspensions, confirming the absence of a link between mortality and suspension age. Types of variants Seven types of somaclonal variants were found (Table 2, Figure 1). Apart from the Multi-stem variant (Figure 1H), the variants corresponded to the descriptions of mutations seen in seed progenies by Cramer (1913) and Krug et al. (1939). The Dwarf (Figure 1C) and Thick leaf (Bullata) (Figures 1F and 1G) phenotypes were recently observed in C. arabica plants derived from embryogenic suspensions (Etienne and Bertrand 2001). Histological markers of these variants, e.g., stomatal density or chloroplast number per guard cell, were observed. We found that the Dwarf, Thick leaf and Angustifolia (Figure 1E) phenotypes were easily characterized by leaf shape (Table 2). Compared to the controls, somaclonal variants had reduced vigor and productivity, with the exception of the Juvenile leaf color (Figure 1B) and Giant (Figure 1D) phenotypes. The Giant and Thick leaf variants were taller than controls. Apical dominance was weak in the Multi-stem variant, which from the juvenile stage was highly branched with two to four stems from cauline buds (Figure 1H). It did not survive in the field in full sunlight, and in shade it grew slowly, producing large leaves but few fruits. The frequency of Variegata, Giant, Thick leaf and Juvenile leaf color variants did not exceed 0.3% (Table 3). The Dwarf variant, which affects the size and productivity of the tree, exceeded 1%. The frequency of the Multi-stem and Angustifolia variants was 2 and 4%, respectively. Effect of cell suspension age on variant frequency Plants derived from 0- and 3-month-old cell suspensions had similar proportions of somaclonal variants (Table 3). However, the proportion of variants among plants from cell cultures older than 6 months increased exponentially with suspension age, according to the formula f = 0.99e 0.267t (r 2 = 0.99). Table 1. Mortality (%) of different genotypes of C. arabica trees regenerated by somatic embryogenesis from embryogenic suspensions of different ages. Observations were made on surviving plants after 18 months in the nursery and 24 months in the field. Clone Suspension age (months) 3σ Confidence 0 3 6 9 12 All interval Control 2.66 4.0 5.30 25.3 9.43 8.52 (4.70, 12.33) H1 5.30 0.86 7.50 6.71 0.47 3.17 (1.30, 5.04) H2 5.74 3.75 9.16 9.75 0 5.08 (2.82, 8.05) H3 11.46 6.93 2.43 16.99 3.25 7.64 (4.98, 10.29) H4 7.40 2.85 6.25 14.40 1.06 6.53 (3.37, 9.68) H5 16.03 9.72 0 6.25 0 7.89 (3.51, 12.26) All 8.52 4.00 5.8 13.70 2.46 6.27 3σ Confidence (5.33, 11.70) (2.03, 5.96) (2.54, 9.05) (9.37, 0.180) (1.17, 4.42) (5.05, 7.48) interval TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

422 ETIENNE AND BERTRAND Table 2. Morphology and productivity of seven C. arabica somaclonal variants. Observations for each variant were made on one or two genotypes in which that variant was the most frequent. Quantitative data were compared with those obtained from 10 normal trees (control). Productivity for the first yield of the variants and of the controls was measured in grams of fresh berries per tree. The values of leaf length and leaf width represent the mean of each tree. Tree height was measured on all variants in the studied genotypes. The significance of differences between variants and controls was evaluated by analysis of variance. Means followed by the same suffix are not significantly different (P < 0.05). Variant Clones observed Phenotype Measurement Value Effect of genotype on variant frequency Variant frequencies differed significantly between Clones H4 and H5, which had over 18% of variants, and Clones H1 and H3, which had around 5% of variants (Table 4). Clone H2, with about 9% of variants, was intermediate between these two groups. However, the differences between clones only appeared in plants derived from cell suspensions that were 6 months old or older. In plants derived from 6-month old cell suspensions, somaclonal variations were 10% for Clones H2 and H3, whereas the proportion of variants for Clones H1, H4 and H5 was similar to that for plants derived from 0- and 3-month-old cell suspensions. For plants derived from the 9-month-old cell suspension, the proportions of variants ranged from 6.25% for Clone H5 to 14.63% for Clone H2, but 2 the differences between clones were not significant (χ obs = 2 5.42 and χ 0.95 = 9.49). For plants derived from the 12-month-old cell suspension, the number of variants produced differed statistically (P < 0.0001) between Clones H4 and H5, which had 95 and 83% of variants, respectively, and Clones H1, H2 and H3, which had 8.53, 18 and 5.91% of variants, respectively. Effect of genotype on variant type Based on growth and productivity loss, somaclonal variants ranked as follows: Multi-stem > Angustifolia > Variegata > Thick leaf > Dwarf > Giant > Juvenile leaf color. The frequency of variant types ranged from about 0.1% to about 10% (Table 5). The Juvenile leaf color variant was mostly restricted to Clone H3. The Dwarf variant occurred in all genotypes but was most frequent in Clones H1, H2 and H3. Multi-stem variants were rare in Clones H1, H2 and H3, but occurred with a frequency of close to 10% in Clones H4 and H5. The Angustifolia variant was the most common in all clones except Clone H3. Within a genotype, the types of variants observed depended on the age of the cell suspension. With increasing cell suspension age, there was a trend toward more serious somaclonal variations in all genotypes. For example, the Multi-stem and Angustifolia variants were found mainly in plants derived from 9- and 12-month-old cell suspensions. Discussion Variant Control Variegata H3 and H4 Variegated leaves, Productivity (g plant 1 ) 484.5 b 1298 a decreased tree vigor Juvenile leaf color H3 Developing leaves changed Productivity (g plant 1 ) 1054 a 1098 a from green to bronze in color, tree vigor unaffected Dwarf H2 Small leaves, small trees Leaf length (cm) 10.40 b 13.57 a Leaf width (cm) 4.45 b 5.89 a Productivity (g plant 1 ) 825 b 1179 a Giant H2 and H4 Normal leaves, taller trees Tree height (cm) 260 a 195 b Productivity (g plant 1 ) 1268 a 1324 a Angustifolia H1 Elongated leaves, fewer or Leaf width (cm) 6.50 b 7.23 a no domatia, longer Number of domatia (leaf 1 ) 3.72 b 10.4 a internodes, taller trees, Tree height (cm) 253 a 190 b decreased tree vigor Productivity (g plant 1 ) 323 b 1245 a Thick leaf (Bullata) H1 and H2 Rounded, lusterless and Leaf length (cm) 13.21 b 14.37 a thick leaves, starry flowers, Leaf width (cm) 7.85 a 6.70 b large-sized fruits Leaf width/leaf length 1.69 b 2.15 a Productivity (g plant 1 ) 678 b 1198 a Multi-stem H4 Highly branched, died in Number of stems emerging 2 to 4 1 the field from cauline bud Productivity (g plant 1 ) 0 b 1590 a Factors affecting variant frequency The proportion of somaclonal variants increases with the number of multiplication cycles, or with the length of culture time (Lörz and Scowcroft 1983, Preil 1986, Cassells and Morrish 1987, Benzion and Phillips 1988, Hartmann et al. 1989, Wang et al. 1992, Yang et al. 1999). Similar findings have been reported for some somatic embryogenesis procedures (Morrish et al. 1983, Symillides et al. 1995, Henry et al. 1996). The existence of somaclonal variants in the progeny of coffee plants propagated by somatic embryogenesis has been documented (Söndähl and Lauritis 1992, Etienne and Bertrand 2001) and TREE PHYSIOLOGY VOLUME 23, 2003

SOMACLONAL VARIATIONS IN COFFEE CELL SUSPENSIONS the effect of genotype demonstrated (Etienne and Bertrand 2001). However, the effects of other culture parameters such as growth regulators and culture age on somaclonal variation have not been established. In our study, we found that the age of C. arabica embryogenic suspensions affected variant frequency. For all genotypes studied, the frequency of variants increased exponentially with suspension age, indicating that, for true-to-type multiplication, it is essential to restrict embryogenic material multiplication times to less than 6 months. However, the initial stages of cell dedifferentiation and embryogenic callus induction were also found to be mutagenic, because variants were regenerated from callus of most of the genotypes. Similar findings have been reported for coffee (Söndahl and Bragin 1991), tomato (Ramulu 1991) and banana, where mutagenesis was found to be associated with excessive use of auxin analogs (Shchukin et al. 1997). As was demonstrated in barley cultures (Ziauddin and Kasha 1990), it is possible that the culture-age effect is associated with prolonged exposure to 2,4-D (1 mg l 1). One strategy for minimizing the occurrence of somaclonal variations might be to change the hormonal composition of the medium (Karp 1991), or to eliminate the use of 2,4-D during the suspension prolifer- TREE PHYSIOLOGY ONLINE at http://heronpublishing.com Figure 1. Characteristics of the seven somaclonal variations observed in Coffea arabica plants derived from embryogenic cell suspensions. Arrows indicate the variant plant material. (A) Variegata variant. (B) Juvenile leaf color variant (bronze normal leaves on the right and green variant leaves on the left). (C) Dwarf variant characterized by a compact phenotype and small leaves. (D) Giant variant in the field. (E) Angustifolia variant (on the right), with elongated leaves. (F) The arrow indicates a branch of the Thick leaf (Bullata) variant bearing few fruits of large size. A branch from a normal plant is shown in the background. (G) On the right, the arrow indicates the rounded, lusterless and thick leaves of the Thick leaf variant. On the left, the normal leaves of the same hybrid. (H) Multi-stem variant in the nursery showing the emergence of four stems from the cauline bud. 423

424 ETIENNE AND BERTRAND Table 3. Number of somaclonal variants in C. arabica somatic embryo-derived trees from embryogenic suspensions of different ages. Observations were made on surviving plants after 18 months in the nursery and 24 months in the field. Variant Suspension age (months) Variant 3σ frequency Confidence 0 3 6 9 12 (%) interval Variegata 1 0 1 0 1 0.09 (0.00, 0.27) Juvenile leaf color 2 0 1 3 9 0.48 (0.05, 0.75) Dwarf 5 3 21 6 7 1.36 (0.72, 2.00) Giant 0 1 1 0 0 0.06 (0.00, 0.20) Angustifolia 0 0 2 37 110 4.83 (3.66, 5.99) Thick leaf (Bullata) 0 2 0 0 2 0.13 (0.00, 0.25) Multi-stem 0 4 0 6 79 2.88 (1.96, 3.62) Number of trees 617 757 390 494 826 3084 Proportion of 1.29 1.32 6.41 10.52 25.18 variants (%) 3σ Confidence (0.0, 2.66) (0.09, 2.68) (2.74, 10.08) (6.41, 14.62) (21.2, 30.3) interval ation stage. Zamarippa et al. (1991) succeeded in achieving long-term multiplication of embryogenic suspensions of C. canephora in the presence of BAP only. A genotypic effect on somaclonal variation in C. arabica has been shown previously in F 1 hybrids (Etienne and Bertrand 2001) and in nine widely cultivated varieties (Söndahl and Bragin 1991). Söndahl and Bragin (1991) showed that plants propagated by somatic embryogenesis, without an embryogenic suspension proliferation stage, included variants with frequencies ranging from 3 to 39%. All of the genotypes in our study exhibited somaclonal variation but there were differences among genotypes in the frequency of variants. Hybrids belonging to the same family, such as H2 and H3 (T8667 Rume Sudan) and H4 and H5 (T5296 Rume Sudan), performed similarly with respect to variant frequency, confirming that genotype influences somaclonal variation. Factors affecting the typology of the variants In the seven somaclonal variants identified, productivity and vegetative vigor were less than or no better than (e.g., Giant and Juvenile leaf color) in seedling control plants. As previously observed (Etienne and Bertrand 2001), we found no variants associated with agronomically interesting quantitative and physiological characteristics, unlike Söndahl and Bragin (1991) who reported the existence of somaclonal variations causing earlier or later fruit ripening. We found that both genotypic factors and culture age affected variant frequency and type. Only Clones H2 and H3 produced the Juvenile leaf color variant, whereas only Clones H4 and H5 produced significant numbers of the Multi-stem variant. Even variants common to all genotypes differed in frequency depending on genetic factors. For example, the Dwarf variants was frequent only in Clones H1, H2 and H3, whereas the Angustifolia variant, which was also common to all clones, was most frequent in Clones H4 and H5. Genotypic control of the typology of somaclonal variants in tomato (Lycopersicon spp.) plants regenerated by in vitro culture has been demonstrated (see review by Ramulu 1991). Moreover, variability in variant types is greater in plants regenerated from wild species (L. peruvianum Mill, L. chilense Dun.) than in plants regenerated from cultivated species (L. esculentum Mill), probably because of a large degree of heterozygosity in the wild species. Table 4. Proportion of somaclonal variants in different genotypes of C. arabica somatic embryo-derived trees from embryogenic suspensions of different ages. Observations were made on surviving plants after 18 months in the nursery and 24 months in the field. Symbols: X = number of variants; and n = number of trees. Clone Suspension age (months) Number Variant 3σ of trees frequency Confidence 0 3 6 9 12 (%) interval X/n (%) X/n (%) X/n (%) X/n (%) X/n (%) H1 3/132 2.3 0/231 0 2/80 2.5 15/134 11.2 18/211 8.5 788 4.82 (2.55, 7.08) H2 2/87 2.3 5/133 3.8 12/120 12 6/41 14.6 20/111 18 492 8.94 (5.20, 12.80) H3 3/157 1.9 1/173 0.6 12/82 14.6 11/153 7.2 20/338 5.9 903 5.09 (2.81, 7.19) H4 1/135 0.6 3/140 2.1 1/64 15.6 17/118 14.4 90/94 95.7 551 20.32 (15.1, 25.13) H5 0/106 0 1/80 1.3 0/44 0 3/48 6.3 60/72 83.3 350 18.28 (11.80, 24.20) TREE PHYSIOLOGY VOLUME 23, 2003

SOMACLONAL VARIATIONS IN COFFEE CELL SUSPENSIONS 425 Table 5. Variant frequency (%) in C. arabica somatic embryo-derived trees for Clones H1 to H5. Observations made on surviving plants after 18 months in the nursery and 24 months in the field. Variant H1 H2 H3 H4 H5 Variegata 0.22 0.18 Juvenile leaf color 0.13 0.4 1.43 Dwarf 1.14 3.05 1.77 0.18 0.28 Giant 0.2 0.18 Angustifolia 3.17 5.28 1.56 9.62 9.12 Thick leaf (Bullata) 0.25 0.2 0.28 Multi-stem 0.13 0.22 10.16 8.57 Number of trees 788 492 903 551 350 We conclude that in C. arabica somaclonal variations are associated with long cell proliferation times. We note that somaclonal variations have also been linked to excessive cell multiplication rates during disorganized growth processes, such as meristem formation or callus multiplication (Karp 1992). Acknowledgments We thank William Solano, Alexis Pereira and Walter Ramirez for their technical assistance. This work received financial and technical backing from PROMECAFE, CATIE and the French Regional Delegation of the Ministry of Foreign Affairs in Costa Rica. References Benzion, B. and R.L. Phillips. 1988. Cytogenetic stability of maize tissue cultures: a cell line pedigree analysis. Genome 30:318 325. Bertrand, B., G. Aguilar, R. Santacreo and F. Anzueto. 1999. El mejoramiento genético en América Central. In Desafios de la Caficultura Centroamericana. Eds. B. Bertrand and B. Rapidel. IICA Publishers, San José, Costa Rica, pp 407 456. Buiatti, M. and F. Gimelli. 1993. Somaclonal variation in Ornamentals. In Creating Genetic Variation in Ornamentals. Proc. XVIIth Eucarpia Symposium, San Remo, Italy, pp 5 24. Cassells, A.C. and F.M. Morrish. 1987. Variation in adventitious regenerants of Begonia-rex Putz. Lucille Closon as a consequence of cell ontogeny callus ageing and frequency of subculture. Sci. Hortic. 32:135 144. Cramer, P.J.S. 1913. Gegevens over de variabiliteit van de in Neederlands Indie verbouwde koffie soorten. Ed. G. Kolff. Batavia, Dept. Landbouw Meded, 696 p. Duncan, R.R. 1997. Tissue culture-induced variation and crop improvement. Adv. Agron. 58:201 240. Etienne, H. and B. Bertrand. 2001. Trueness-to-type and agronomic characteristics of Coffea arabica trees micropropagated by the embryogenic cell suspension technique. Tree Physiol. 21: 1031 1038. Etienne, H., B. Bertrand, F. Anthony, F. Côte and M. Berthouly. 1997. Somatic embryogenesis: a tool for coffee genetic improvement. In Proc. 17th Colloquium International Coffee Science Association, Nairobi, Kenya. ASIC Publishers, Vevey, Switzerland, pp 457 465. Etienne, H., F. Anthony, S. Dussert, D. Fernandez, P. Lashermes and B. Bertrand. 2002. Biotechnological applications for the improvement of coffee (Coffea arabica L.). In Vitro Cell. Dev. Biol.-Plant 38:129 138. Etienne-Barry, D., B. Bertrand, N. Vasquez and H. Etienne. 1999. Direct sowing of Coffea arabica somatic embryos mass-produced in a bioreactor and regeneration of plants. Plant. Cell. Rep. 19: 111 117. Hartmann, C., Y. Henry, J. de Buyser, C. Aubry and A. Rode. 1989. Identification of new mitochondrial genome organizations in wheat plants regenerated from somatic tissue cultures. Theor. Appl. Genet. 77:169 175. Henry, Y., J.-L. Marcotte and J. de Buyser. 1996. The effects of aneuploidy on karyotype abnormalities in wheat plants regenerated from short- and long-term somatic embryogenesis. Plant Sci. 114: 101 109. Karp, A. 1991. On the current understanding of somaclonal variation. In Oxford Surveys of Plant Molecular Cell Biology, Vol. 7. Ed. B.J. Miflin. Oxford University Press, London, pp 1 58. Karp, A. 1992. The effects of plant growth regulators on somaclonal variation in plants regenerated from tissue cultures. Annu. Bull. Br. Soc. Plant Growth Regul. 2:1 9. Krug, C.A., J.E.T. Mendes and A. Carvalho. 1939. Taxonomia de Coffea arabica L. Instituto Agronomico do Estado, em Campinas. Boletím técnico 62:1 57. Lörz, H., W.R. Scowcroft. 1983. Variability among plants and their progeny regenerated from protoplast of Su/su heterozygotes of Nicotiana tabacum. Theor. Appl. Genet. 66:67 75. Morrish, F.M., W.W. Hanna and I.K. Vasil. 1983. The expression and perpetuation of inherent somatic variation in regenerants from embryogenic cultures of Pennisetum glaucum (L.) R. Br. (pearl millet). Theor. Appl. Genet. 80:409 416. Preil, W. 1986. In vitro propagation and breeding of ornamental plants: advantages and disadvantages of variability. In Genetic Manipulation in Plant Breeding. Ed. Walter de Gruyter, Berlin, pp 337 403. Ramulu, K.S. 1991. Genetic variation in in-vitro cultures and regenerated plants in tomato and its implications. In Monographs on Theoretical and Applied Genetics, Vol. 14. Genetic Improvement of Tomato. Ed. P. Kalloo. Springer-Verlag, Heidelberg, pp 259 275. Shchukin, A., D. Ben-Bassat and Y. Israeli. 1997. Plant regeneration via somatic embryogenesis in grand nain banana and its effect on somaclonal variation. Acta Hortic. 447:317 318. Söndahl, M.R. and A. Bragin. 1991. Somaclonal variation as a breeding tool for coffee improvement. In Proc. 14th Colloquium International Coffee Science Association, San Francisco, USA. ASIC Publishers, Vevey, Switzerland, pp 701 710. Söndahl, M.R. and J.A. Lauritis. 1992. Coffee. In Biotechnology of Perennial Fruit Crops. Eds. F.A. Hammerschlag and R.E. Litz. CAB International, Wallingford, UK 8:401 420. TREE PHYSIOLOGY ONLINE at http://heronpublishing.com

426 ETIENNE AND BERTRAND Starisky, G. and G.A.M. Van Hasselt. 1980. The synchronised mass propagation of Coffea canephora in vitro. In Proc. 9th Colloquium International Coffee Science Association, London, England. ASIC Publishers, Vevey, Switzerland, pp 597 602. Symillides, Y., Y. Henry and J. de Buyser. 1995. Analysis of Chinese Spring regenerants obtained from short- and long-term wheat somatic embryogenesis. Euphytica 82:263 268. Van Boxtel, J. and M. Berthouly. 1996. High frequency somatic embryogenesis in coffee leaves. Factors influencing embryogenesis, and subsequent proliferation and regeneration in liquid medium. Plant Cell Tissue Organ Cult. 44:7 17. Wang, X.H., P.A. Lazzeri and H. Lörz. 1992. Chromosomal variation in dividing protoplasts derived from cell suspensions of barley (Hordeum vulgare L.). Theor. Appl. Genet. 85:181 185. Yang, H., Y. Tabei, H. Kamada, T. Kayano and F. Takaiwa. 1999. Detection of somaclonal variation in tissue cultured rice cells using digoxigenin-based random amplified polymorphic DNA. Plant. Cell. Rep. 18:520 526. Zamarripa, A., J.P. Ducos, H. Bollon, M. Dufour and V. Pétiard. 1991. Production of coffee somatic embryos in liquid medium: effects of inoculation density and renewal of the medium. Café Cacao Thé 35:233 244. Ziauddin, A. and K.J. Kasha. 1990. Long-term callus cultures of diploid barley (Hordeum vulgare L.). II. Effects of auxins on chromosomal status of cultures and regeneration of plants. Euphytica 48:279 286. TREE PHYSIOLOGY VOLUME 23, 2003