Six clones of sweet potato (Ipomoea batatas Lam.; Convolvulaceae) were used. Five clones, cultivars Quangshu, Zho, 953, 865 and 90, were obtained from the Xuzhou Sweet Potato Research Centre, China, and one clone, cv. Duclos 11, from the ‘Institut national de la recherche agronomique’ (INRA) of the French West Indies. The plant materials were maintained and multiplied in vitro by subculture of leafy node-cuttings, at intervals of six weeks, on basal MS medium [18], supplemented with vitamins [19], 20 g L⁻¹ sucrose, 0.5 mg L⁻¹ IAA and solidified with 7 g L⁻¹ agar [17]. The cultures were kept in a growth chamber at 27±2°C with 16 h day⁻¹ at 62 μM m⁻² s⁻¹, and 60% humidity.

For induction of somatic embryogenesis, lateral buds with 2 to 4 leaf primordia were taken from the 10 to 12 nodes from apices of 7–8-week-old plants with a sterile hypodermic needle and under a binocular microscope. About 10 buds were placed on the solidified induction medium contained in a 10×60-mm Petri dish. The medium was composed of basal MS medium, vitamins [19], 30 g L⁻¹ sucrose and supplemented with one of the three different auxins, 2,4-D, 2,4,5-T used at two concentrations, 5 and 10 μM, and Picloram at only 10 μM. The growth regulators were added to the medium before autoclaving. The medium was solidified with either 7 g L⁻¹ Agar (Bacto Agar – DIFCO) or 3 g L⁻¹ Gelrite. The pH was adjusted to 5.8. The induction was carried out in darkness at 27±2°C. Embryogenic calli were then divided into fragments of 2–3-mm diameter, and subcultured on the multiplication medium with the same composition as for induction. Further development into embryos and their conversion into plants required the transfer of embryogenic calli onto either hormone-free medium or supplemented with combination of growth regulators, particularly ABA (Sigma) for embryo maturation. For callus multiplication and plant regeneration, the cultures were kept in the growth chamber with the same conditions of temperature and illumination as for plant micropropagation, i.e. 27±2°C with 16 h day⁻¹ at 62 μM m⁻² s⁻¹, and 60% humidity.

Among hundreds of plants derived from somatic embryos of cv. Zho, a sample of 50 clones was randomly selected and multiplied for the study of the genetic stability. Flow cytometry was used to quantify DNA for determination of the ploidy level, as described by Fock et al. [20]. Briefly, about 1 cm² of leaf material from in vitro plants was chopped with a razor blade in 1 ml of a buffer solution containing CPW salts [21], 0.5 M mannitol, 0.25% (w/v) PEG, 0.5% (w/v) Triton X-100, 0.25% (v/v) mercaptoethanol at pH 6.5–7.0. Crude samples were filtered through a 40-μm mesh nylon and stained with 4,6 diaminido-2-phenylindol (DAPI, 5 μg ml⁻¹). DNA analysis was performed on a PARTEC CA II flow cytometer (Chemunex, Maison-Alfort, France) equipped with a 100-W mercury lamp (type HBO). Blue fluorescence at 455 nm was recorded as a function of the DNA content. Because of their exceptional stability and regularity, murine resting splenic B lymphocytes were used to calibrate the fluorescence scale. The external references of the hexaploid status was provided by the use of the hexaploid (6X) parental plants of sweet potato (2n=6x=96 chromosomes) [3]. The DNA distribution was analysed, by using DPAC software, on histograms generated from at least 10,000 nuclei.

Each experiment of induction of somatic embryogenesis involving the combination of factors was repeated at least six times, using a total of nearly 4000 lateral buds. The number and percentage of embryogenic responses were collected 10 weeks after induction. The data, expressed in frequency (x), were transformed into arcsinx (angular transformation) and the frequency zero into 1/(4n) [22], n being the number of explants. The transformed variates were subjected to statistical analysis, using a fixed model of analysis of variance (ANOVA) with one or two criteria of cross classification. Duncan's multiple-range test [23] was used for means separation.

The lateral buds (Fig. 1A) swelled slightly after two weeks of culture in the induction medium. They lost their compact consistency and 1.5–10.7% of cultured buds, depending on the medium and variety, rise to a mucilaginous milky mass in which compact embryogenic calli developed from some explants (Fig. 1B), while most of them became either a small callus or friable fast-growing non-embryogenic callus. Generally, the small callus rapidly turned brown and died within a few days. The fast-growing friable callus was white to brown in colour, and composed of large translucent cells. It never gave rise to embryogenic structures. The compact embryogenic callus was yellowish or white and comprised pink or red areas, due to the presence of anthocyanin, depending on the genotype, particularly cv. Zho (Fig. 1C and D). It was composed of very tightly packed isodiametric cells. In the regeneration medium, embryos appeared; they were observed at different stages (Fig. 1E). These bipolar structures developed root before activation of stem meristem apex, leading to complete plants (Fig. 1F).

The effect of two types of gelling agent, Agar and Gelrite, was assessed for induction of embryogenic response in six cultivars of sweet potato (Table 1). The ANOVA of the embryogenic response (%) showed very highly significant effects of the gelling agent and genotype. Irrespective of the genotype, the medium solidified with Agar gave the best embryogenic response with 3.53%, while very few lateral buds in Gelrite were induced to form embryogenic callus, yielding only 0.45%. Of the six cultivars screened, only three, i.e. Zho, 865 and 90, gave rise to an embryogenic response (Table 1). The best embryogenic response was obtained from the cultivars Zho and 865, particularly in the medium solidified with Agar and containing a low level of 2,4,5-T (5 μM), giving 10.70 and 14.68% of embryogenic callus, respectively. Despite the non-significance of the interaction effect between the gelling agent and genotype, only 1.46% of lateral buds of cultivar 90 were induced to form the embryogenic callus in the medium containing Gelrite and a high level of 2,4,5-T (10 μM). This low frequency of embryogenic response was not significantly different from zero (Table 1).

Comparison was made between three types of auxin, 2,4-D, 2,4,5-T used at 5 and 10 μM, and Picloram at 10 μM for the assessment of the embryogenic response in six sweet potato cultivars cultured on a medium solidified with Agar (Table 2). The ANOVA of the frequency of embryogenic response showed very highly significant effects of the type of auxin and the genotype, as well as their interaction (Table 2). This indicates that the effect of the auxin type on the embryogenic response varied with the genotype of sweet potato. Surprisingly, only two cultivars, Zho and 856, were induced to form few embryogenic calli in medium with a low level of 2.4-D (5 μM), while a high level of 2,4-D (10 μM) induced no embryogenic response at all (Table 2). On the contrary, best results were obtained in cv. Zho with 2,4,5-T or Picloram, yielding a mean of 10.33% of embryogenic response whatever the concentration used. The genotype 865 gave 14.7% of embryogenic response in 5 μM 2,4,5-T and only 6.59% in 10 μM Picloram (Table 2).

The compact embryogenic callus was multiplied by regular subcultures in the fresh medium used for embryogenic induction at intervals of 2 months. The cultures were kept in darkness. During successive subcultures, the embryogenic callus grew up and developed into a new embryogenic callus. Besides, it also formed a mucilaginous milky mass from which emerged compact embryogenic callus. Nevertheless, after several subcultures, some sectors of compact embryogenic callus became necrosed or friable and non-embryogenic. The non-embryogenic callus had to be carefully removed before each subculture, as it grew twice as fast as the embryogenic callus and rapidly invaded all the cultures. Moreover, it could never be induced to form embryogenic callus again, thus losing the ability to regenerate plants. Better proliferation of embryogenic than of non-embryogenic callus could selectively be enhanced on the multiplication medium containing 5 μM 2,4,5-T or the combination of 10 μM 2,4-D and 1 μM BAP or kinetin. By applying this procedure, embryogenic calli of sweet potato, with sustained ability to further regenerate plants, have regularly been maintained for several years.

For further differentiation of embryos, the compact embryogenic calli subcultured on the induction or multiplication medium were illuminated as for plant propagation. After four weeks in the light, embryos at globular stage were developed, but their further development into advanced stages was achieved upon their transfer to hormone-free medium. However, the maturation of embryos and particularly their conversion to plants was dramatically improved by subculture on medium with the combination of 1 μM 2,4-D and 1 μM Kinetin before transfer of mature embryos onto hormone-free medium for further plant development. ABA alone was also tested at 2.5, 5, 10 and 15 μM concentrations (data not shown), and the optimal maturation of embryos was obtained with 5 μM. Those treatments have resulted in successful regeneration of 5–10 plants per cluster of embryogenic callus within less than two months. The embryo-derived plants displayed apparently normal morphology. They rooted well and grew vigorously when subcultured on basal MS medium supplemented with 0.5 mg L⁻¹ IAA for multiplication.

The ploidy level of regenerated plants was determined by comparing the position of dominant peaks corresponding to nuclei at G0–G1 phases of the cell cycle, between embryo-derived plants and parental lines. The analysis showed that the dominant peak of the hexaploid parental line, cv. Zho, was located at channel 83, and that of the sample of 50 embryo-derived clones, which had been examined, ranged from channels 80 to 85, giving a mean value of 82.5±0.6 (Table 3). This mean value was practically identical to that of the control plants as the values of the coefficient of variation were relatively high, ranging from 4.3 to 6.5%. Consequently, the ploidy level of embryo-derived plants could be considered identical to that of the original hexaploid (6x) plants of sweet potato.