Shoots were transferred to the multiplication culture medium (as described above except: 9.3 μM 6-benziladenine and 1.6 μM naphthalene-acetic acid). They were subcultured at 42 d intervals for 168 d. Twenty-four thousand seven hundred and sixty-eight shoots were then obtained.

Three hundred young leaves were randomly selected as explants for callus formation. The culture medium included: Murashige and Skoog salts [46], 100 mg·l^–1 myoinositol, 0.1 mg·l^–1 thiamine-HCl, 30 g·l^–1 sucrose, 29.0 μM naphthalene-acetic acid and 9.7 μM kinetin. The calli were proliferated for 4 months with subcultures every 30 d.

Five hundred calli (Ø: 3 mm) were randomly selected and transferred to the plantlet regeneration medium: Murashige and Skoog salts [46], 100 mg·l^–1 myo-inositol, 0.1 mg·l^–1 thiamine-HCl, 30 g·l^–1 sucrose, 0.9 μM indole- 3-butyric acid, 1.1 μM 6-benzyladenine and 0.09 μM gibberellic acid. Four hundred and twenty-seven in vitroplantlets were obtained and later hardened in ex vitro for 6 months [47].

For ex vitro hardening, plantlets were placed in plastic trays containing 82 cm³ of a mixture zeolite + filter cake (1:1). Microject automated irrigations for 25 s every 30 min were applied. Plantlets were kept under a photosynthetic photon flux density of 458 μmol·m^–2·s^–1. Standard phytosanitary controls were applied. After hardening of in vitro-plantlets, 387 plantlets were transferred to the field environment and asexually propagated for two generations (30 months). The donor cultivar was used as a control. Two phenotype variants were then identified: P3R5 and Dwarf. A more detailed study was carried out to compare these two variants with the donor plant (cv. Red Spanish Pinar, plant material never cultured in vitro). The experiment was developed in the Field Experimental Station at the Bioplant Centre. A random block design was implemented (80 plants/clone). Field management was performed according to instructions recommended by the Cuban Ministry for Agriculture. The most important pineapple phenotypic traits were recorded during 18 months. The agricultural evaluations were made in field conditions.

In the second generation, in order to perform the AFLP characterization, young leaves of the donor genotype (cv. Red Spanish Pinar) and the two variants (P3R5 and Dwarf) were collected. Samples were stored at –20˚C until DNA extraction. Extraction started from 250 mg fresh mass that were finely grounded in liquid nitrogen. Extraction buffer (650 μl) was then added. It included: Tris-Cl (pH 7.5, 50 mmol·l^–1), ethylene-diamine-tetraacetic acid (20 mmol·l^–1), sodium chloride (0.3 mmol·l^–1), sarcosil (2%), sodium dodecyl sulfate (0.5%) and urea (4.8 mmol·l^–1). A mixture (650 μl) of phenol:chloroform: isoamyl alcohol (25:24:1, v:v:v) was added. Samples were centrifuged for 15 min at 12,000 rpm at room temperature. Supernatants were collected and the pellets, discarded. Isopropanol (0.8 volumes) was added. Samples were shortly shaken in a vortex and incubated for 10 min at room temperature. Samples were centrifuged for 10 min at 12,000 rpm. Supernatant was discarded and the pellet was washed with ethanol (70%). DNA was dried (vacuum) and dissolved in 50 μl water supplemented with 10 μg·ml^–1 RNase A. DNA integrity and purity was checked by electrophoresis in agarose gels (0.8%) stained with ethidium bromide. Concentration was estimated visually by comparison with standards (100 - 1000 ng·μl^–1). Concentrations of DNA samples were adjusted to 500 ng·μl^–1.

AFLP technique was carried out [48]. The digestion of genome DNA, the pre-amplification with a selective base and the selective amplification was performed [47]. Autoradiographs were analyzed visually to build a dicotomic numerical matrix: the number one was assigned when the band was present while zero was assigned when absent. We disregarded all weak and low peak AFLP bands. The matrix was processed with the NTSYSpc software [49]. The simple matching index was used to create a matrix of similarity. From this matrix, a matrix of genetic distance was obtained. The UPGMA (Unweighted Pair Group Method with Arithmetic Averages) method was used to generate a dendogram.

In a subsequent procedure, the three plant materials were transferred to the Pineapple Germplasm Bank at the Bioplant Centre in a random block design. Plants grew for 6 months and then D leaves [50] of Red Spanish Pinar (donor), P3R5 and Dwarf were collected. Ten plants per genotype were studied (one leaf per plant). The stoma diameter, number of stomata per mm², diameter of leaf vascular tissue, thickness of the leaf aquiferous parenchyma, and thickness of the leaf photosynthetic parenchyma were measured [51]. The photosynthetic rate, the transpiration rate, the water use efficiency, and the internal leaf CO₂ concentration were recorded using a Portable CIRAS-2 Photosynthesis System (Europe, PP Systems, UK); covering with the leaf, the whole area of the cuvette (PLC6, 2.5 cm²). The carbon dioxide concentration and the relative humidity of the air entering the cuvette were 375 μmol·mol⁻¹ and 80% respectively, under environmental temperature (25˚C - 27˚C). Prior to obtaining the experimental data, we measured the maximum light intensity at which photosynthesis was stable which was attained at 600 μmol·m⁻²·s⁻¹.

To determine the levels of chlorophyll pigments (a, b, total), leaves were thinly grounded in liquid nitrogen. Evaluations were made [52]. Extraction was carried out with 5.0 ml acetone (80%, v:v). The samples were centrifuged (12,000 rpm, 4˚C, 15 min), supernatants collected and absorbances at 647 and 664 nm were recorded.

Contents of malondialdehyde and other aldehydes [53]; and phenolics (cell wall-linked, free, and total were determined [54]. Total protein contents were recorded [55]. Enzymatic activities and specific activities of phenylalanine ammonia-lyase [56], superoxide dismutase [57], and phosphoenol pyruvate carboxylase [58,59] were also measured.

The Statistical Package for Social Sciences (Version 8.0 for Windows, SPSS Inc.) was used to perform One-Way ANOVA and Tukey tests (P ≤ 0.05). The Euclidean distances of each somaclonal variant to the donor plant material were calculated. Data were standardized to vary from 0 to 1 [60].

Only two plant materials were found to be “solid” somaclonal variants after studying three vegetative generations (P3R5 and Dwarf) and the AFLP analysis, which represents 0.52% (2 somaclonal variants/387 plants transferred to the field). The dendogram generated with the AFLP information revealed an existing genetic distance among the somaclonal variants and the donor plant [43]. The genetic distances among the three plant materials are not too significant. However, as they have different banding patterns, they are different at the genetic level.

The agricultural characterization of the third vegetative generation in Tables 1(a)-(b) and Figure 1 shows that, in comparison with the donor plant (cv. Red Spanish Pinar), the variant P3R5 showed differences in the number of slips and suckers, and in the presence of thorns in the leaves and in the fruit crowns. The somaclonal variant Dwarf was different from the donor plant in regard to the plant height; the peduncle diameter; the number of shoots, slips and suckers; the fruit mass with crown; the number of eyes in the fruit; the fruit height and diameter; the leaf color; the plant architecture; the length of plant generation cycle; and the fruit color and shape (Table 1(a), agricultural characterization of the third vegetative generation).

The morphological, physiological and biochemical characterization of D leaves in Table 1(b) shows that, in comparison with the donor plant (cv. Red Spanish

Pinar), the variant P3R5 showed statistically significant differences in 15 indicators while Dwarf in 17 variables. Compared to the donor plant, P3R5 somaclonal variant showed significant low values in several aspects, but mainly in the transpiration rate that only reached 28% of the rate in the donor (11.5 mmol H₂O m⁻²·s⁻¹/41.6 mmol H₂O m⁻²·s⁻¹). Moreover, content of free phenolics in P3R5 merely represented 37% (472.58 mg·g⁻¹ fresh leaf mass/1273.44 mg·g⁻¹ fresh leaf mass). Significant increases were also recorded in P3R5 in comparison with cv. Red Spanish Pinar. For instance, the donor showed 39% of the water use efficiency evaluated in P3R5 (0.7 mmol CO₂ mol⁻¹ H₂O/1.8 mmol CO₂ mol⁻¹ H₂O) (Table 1).

Comparing Dwarf somaclonal variant with the donor, it only reached 35% of the thickness of the photosynthetic parenchyma of D leaf recorded in the donor (19.9 µm/57.6 µm), 37% of the thickness of the leaf aquiferous parenchyma (43.8 µm/119.1 µm), and 39% of the superoxide dismutase activity and specific activity (0.33 U·mg⁻¹ fresh leaf mass/0.85 U·mg⁻¹ fresh leaf mass, 8.27 U·mg⁻¹

of protein/21.16 U·mg⁻¹ of protein, respectively). On the other hand, the donor plant material only showed about 46% of the phenylalanine ammonia-lyase activity and specific activity (0.29 U·mg⁻¹ fresh leaf mass/0.62 U·mg⁻¹ fresh leaf mass, 0.0071 U·mg⁻¹ of protein/0.0157 U·mg⁻¹ of protein) (Table 1).

Changes in the above mentioned morphological, physiological and biochemical indicators have been frequently studied when plants have been submitted to different sources of stress. Reference [61], recorded the optimization of CO₂ gain through stomatal aperture while minimizing water loss in rice. The effects of flooding and drought stress on citrus seedlings physiology were measured [62]. The response of cucumber seedlings to drought stress also were measured [63]. However, to our knowledge, the effects of somaclonal variation on plant physiology have not been deeply studied. Further studies are required to elucidate the mechanisms that explain the differences observed in P3R5 and Dwarf somaclonal variants.

The overall coefficients of variation in Table 1 indicate that the number of shoots was significantly different among the three plant materials (173.21%).

Dwarf showed two shoots per plant while P3R5 and the donor did not form any shoot. We classified the overall coefficients of variation of the other phenotype indicators in three categories: less than 23%, between 23 and 46%, and over 46%. Then we observed that water use efficiency, chlorophyll b concentration, total chlorophyll concentration, transpiration rate, chlorophyll a concentration, thickness of the leaf photosynthetic parenchyma, fruit mass with crown, free phenolics content and superoxide dismutase specific activity were also very different among the three plant materials. However, other aldehyde content, malondialdehyde content, content of cell wall-linked phenolics, protein content, phosphoenol pyruvate carboxylase activity, phosphoenol pyruvate carboxylase specific activity, total content of phenolics, number of stomata per mm², stoma diameter, photosynthetic rate, plant generation cycle, fruit content of vitamin C, number of crowns in the fruit and fruit acidity showed low variability.

Table 2 summarizes the phenotypic changes of P3R5 and Dwarf somaclonal variants with respect to the donor plant material. We have used 44 indicators based on a wide range of horticultural and physiological traits. These data clearly show the various aspects where somaclonal variation can occur in pineapple. P3R5 differed from the donor in 19 variables (19/44; 43.18%), while Dwarf in 31 indicators (31/44; 70.45%; Table 2).