Recovery of plants and ploidy level analysis

After pollination of 350 'Clemenules' clementine flowers with irradiated pollen of 'Fortune'mandarin, we obtained 270 fruits containing 1744 seeds approximately 4-5 mm in length, much smaller than normal seeds (10-12 mm on average). Only 2.9% of these seeds contained embryos (Figures 1a and 1b). A total of 51 embryos were cultivated in vitro, 13 of which developed either by direct germination or through the formation of embryogenic calli (Figures 1c, d, e and 1f). To regenerate plants it was necessary to use the technique of shoot tip grafting in vitro [46] because embryos did not develop roots and when they produced roots they were small and very weak. Nine plants were obtained by direct germination of embryos and subsequent in vitro micrografting (Figures 1g and 1h). Four embryogenic calli were also induced (Table 1) producing a total of 96 embryos, from which 16 plants were recovered by in vitro micrografting of resulting shoots.

Callus	Ploidy	N° obtained embryos	N° germinated embryos	N° obtained plants	N° haploid plants	N° diploid plants	N° aneuploid plants
A	Haploid	10	10	10	10	0	0
B	Aneuploid	40	12	3	0	0	3
C	Haploid	44	10	1	0	1	0
D	Haploid	2	2	2	2	0	0

Ploidy level was initially evaluated by flow cytometry. Eight of the nine plants obtained by direct germination of the embryos were haploid (Figure 2a) and one was diploid. The ploidy level of three of the four calli obtained (Table 1) was haploid, whereas one (callus B) was suspected to be aneuploid. The twelve plants obtained from haploid calli A and D were haploid. One diploid plant was obtained from haploid callus C, whereas we regenerated three plants with probable aneuploidy from callus B (Figure 2b). Seven haploid plants and the diploid plant from direct germination were very weak and died before making other characterizations.

Noteworthy, one of the propagations of the haploid plant G produced a branch with larger and wider leaves than those of the rest of the plant. The ploidy level of all leaves pertaining to this branch was determined by flow cytometry. Both diploid and haploid leaves were identified. All buds corresponding to the leaves that displayed diploid profiles were grafted in the greenhouse onto a vigorous rooststock. When buds sprouted and the leaves were completely formed, we again determined the ploidy level. Using this method, a diploid plant, arising from in vivo spontaneous somatic duplication of the chromosome number of the haploid line G was obtained and confirmed by chromosome counts.

Chromosome counts were done on three lines and confirmed that haploid plant G had nine chromosomes (2n = x = 9) (Figure 3a), the diploid plant obtained from callus C had eighteen chromosomes (2n = 2x = 18) (Figure 3b) and we confirmed that the plants arising from callus B were aneuploid with twenty-two chromosomes (2n = 2x = 22) (Figure 3c).

SSR analysis of plants obtained

All haploid, diploid and aneuploid plants established in the greenhouse, together with diploid 'Clemenules' clementine and 'Fortune' mandarin (the genotype used for irradiated pollen), were analysed with five SSR markers, heterozygotic in clementine. For each locus, all the haploid plants and the diploid plant C.1 possessed a single allele. All plants recovered from a same callus were identical for all markers. A restitution of clementine heterozygosity was observed only in the aneuploid plants for the markers Ci03C08, Mest 15 and TAA 15 (Figure 4).

For all SSR markers analysed, the haploid plant G and the diploid plant C.1 possessed only one of the clementine alleles. Therefore, no restitution of maternal heterozygosity occurred for these markers for either the haploid or diploid plant (which, hereafter, is considered a double haploid). The aneuploid plant displayed incomplete restitution of maternal heterozygosity with a heterozygosity percentage of 61.5% concerning all linkage groups. With no specific allele of 'Fortune' mandarin, the aneuploid plants have a very low probability of being hybrids with this mandarin. Indeed taking into account only one marker of each of the six linkage groups with specific alleles the probability is (1/2)⁶ (less than 0.016) The restitution of clementine heterozygosity for markers assigned to all linkage groups of the citrus genetic map indicate that the aneuploid plants could not have arisen from the spontaneous duplication of the chromosome stock of aneuploid callus cells with eleven chromosomes. Moreover, the incomplete restitution of the maternal heterozygosity discounts the hypothesis of somaclonal variation from maternal somatic tissue.

The diploid plant obtained from in vivo spontaneous somatic duplication of the chromosome number of the haploid line G was also confirmed fully homozygous for the same allele as the haploid plant G by using the same 52 SSR markers.

Morphological characterization

There were statistically significant differences in all the variables analysed, according to the different ploidy level (Table 3 and Figure 5). The double haploid genotype had leaves with the greatest average foliar area (25.9 cm²), followed by the diploid 'Clemenules' clementine and the aneuploid genotype (19.4 and 4.0 cm², respectively). The haploid plants had lower values, oscillating between 2.1 and 3.0 cm², depending on the genotype.

Genotype	Ploidy	Average leaf area (cm2)	Average leaf width (cm)	Average leaf length (cm)
Clemenules	Diploid	19.4c	3.3e	10.3e
B.1	Aneuploid	4.0b	1.4c	4.4c
C.1	Double haploid	25.9d	5.3f	7.2d
A.1	Haploid	2.1a	1.1a	3.2b
G	Haploid	3.0ab	1.1a	4.6c
D.1	Haploid	2.6a	1.2ab	3.8b
E	Haploid	2.9a	2.0d	2.3a
F	Haploid	2.3a	1.4bc	2.5a

Different letters in the same column indicate statistical differences at a significance level below 0.0001.

The double haploid plant also had the widest leaves (5.3 cm), followed by the diploid 'Clemenules' clementine and the haploid plant E (3.3 and 2.0 cm, respectively).

With respect to leaf length, the 'Clemenules' clementine had the largest value (10.3 cm), followed by the double haploid plant (7.2 cm). The haploid plants possessed a foliar length that varied between 2.3 and 4.6 cm. The maximum value of the haploid plants was similar to the leaf length of the aneuploid plant (4.4 cm).

Histological characterization

The histological structure of anthers of the haploid plant G is similar to that of the diploid 'Clemenules' clementine (Figures 6a and 6b). Nevertheless, differences were observed in the width, height, percentage of anthers with locules and percentage of locules with pollen grains (Table 4). Values for width and height of the haploid anthers were, respectively, 58% and 64% of corresponding values for diploid plants. Diploid anthers always contained two locules with well-formed pollen grains, whereas only 4.7% of the haploid anthers possessed locules and the pollen grains were malformed.

	Genotypes
	Haploid G	Diploid 'Clemenules'
Anthers
Width (μm)	324.9a	777.9b
Height (μm)	181.4a	504.1b
% anthers with developed locules	23.8a	100b
% anthers locules with pollen grains	4.7a	100b
Ovaries
Diameter (μm)	702.2a	1428.7b
N° of locules per ovary	8a	10b
N° of ovules per section	6.6a	15.2b
Styles
Diameter (μm)	440.6a	938.1b
Stigmas
Diameter (μm)	591.1a	1496.4b

Different letters in the same line indicate statistical differences at a significance level below 0.0001.

In the ovaries of the haploid plant we observed a discontinuity in the central axis, which was not fused with the carpellous leaves (Figures 6c and 6d.). The diameter of haploid ovaries (Table 4) was approximately 50% smaller than those of diploid plants, although both displayed the same external morphology. Haploid ovaries had an average of eight locules per ovary, whereas the diploid plant contained ten locules. Haploid ovaries contained approximately half the number of ovules than diploid ovaries. At the same phenological stage, haploid ovules presented reduced growth compared to diploid plants in that the ovules of the diploid plant were totally developed whereas, for most of the haploid ovules, the inner and outer tegument did not completely surround the nucellus (Figures 6c and 6d).

The histological structures of haploid styles and stigmas were similar to diploid plants (Figures 6e, f, g and 6h). Haploid styles were 46 - 48% smaller than diploid styles and haploid stigmas were approximately 40% smaller than those of diploid plants (Table 4).

Origin of the obtained plants

We have obtained haploid plants (2n = x = 9), aneuploid plants (2n = 2x+4 = 22) and a diploid plant (2n = 2x = 18) of 'Clemenules' clementine by in vitro culture of embryos produced following pollination with irradiated pollen of 'Fortune' mandarin.

The haploid plants originated from gamete reduction of 'Clemenules' clementine and regenerated by direct embryo germination or from embryogenic calli induced from the haploid seed embryos.

Genetic analysis with SSR markers confirmed the gynogenetic origin of all obtained plants, since we did not observe specific alleles of the male parent ('Fortune' mandarin) in any plant. No introgression of pollinator genome fragments, after gynogenesis induced by irradiated pollen, have been observed in citrus and other tree species [37, 39, 45]. However it should be noted that in Nicotiana [47, 48] some introgression of male genome fragments have been identified in regenerated gynogenetic plant and confirmed by morphological markers analysis.

The diploid plant was obtained from an originally haploid callus, in which a spontaneous duplication of the chromosome stock took place. Double haploid [32, 33] and trihaploid citrus plants [26] have previously been obtained by androgenesis. The duplication and triplication of the chromosome stocks of haploid clementine calli have been previously described [49]. Double haploid pummelos (C. grandis (L.) Osb) have also been obtained by colchicine treatment in vivo of axillary buds of haploid plants [50]. This plant is the first double haploid plant of 'Clemenules' clementine obtained by gynogenesis in situ, induced by irradiated pollen.

For the aneuploid plants, we propose that they arise from an aneuploid callus produced by an embryo derived, by gynogenesis, from a non-reduced aneuploid gamete (2n = 2x+4 = 22) of 'Clemenules' clementine. The formation of non-reduced gametes in citrus is a well-known phenomenon and is quite common in oranges and mandarins [51–54]. The abnormal meiosis that generates non-reduced gametes is also favourable for the formation of aneuploid gametes. Aneuploid citrus plants have previously been obtained in 2x × 2x, 2x × 4x, 4x × 2x and 2x × 3x hybridisations [43, 55], but this is the first report of an aneuploid plant obtained by gynogenesis of a non-reduced aneuploid megagametophyte.

Effect of ploidy level on the morphologic characteristics of regenerated plants

Significant differences in leaf morphology and size were observed in relation to ploidy level. A diversity of leaf morphologies was also observed among the haploid plants, obtained from independent meiotic events.

The double haploid plant of 'Clemenules' clementine displayed shorter internodal segments, thorns, a more robust appearance and more vigorous growth than the majority of haploid plants. Nevertheless with time, dieback of branches was observed and periodical propagations were needed to avoid the loss of this genotype. Duplication of the ploidy level (from haploid to double haploid) increased the area, length and width of leaves. The floral organs of haploid clementine were smaller than those of the diploid parental plant. The haploid ovaries, styles and stigmas were approximately 40 - 50% smaller than those of diploid plants. A comparative histological study of the foliar structure of diploids and autotetraploid plants of 'Valencia' sweet orange and 'Femminello' lemon, Romero-Aranda et al., [56] revealed that an increase in ploidy level was associated with an increase in cellular volume, leading to autotetraploid leaves between 20 - 30% thicker than in diploid plants. In plants of Solanum phureja with the same genetic structure but with different levels of ploidy (1x, 2x and 4x), a progressive increase in nuclei and cell volumes was coincident with increasing ploidy. In addition, 2x and 4x plants showed greater development and vigour than 1x plants [57]. Similarly, haploid Vacccinum spp. and apple tree plants also display a very weak appearance and poor growth compared with diploid plants [38, 58].

Plant vigour seems to be particularly affected by haploidy in most citrus species. All haploid plants of clementine, obtained by androgenesis or gynogenesis, displayed a weak appearance and poor growth [25, 31, 40, 44, 45] and most plants typically died in the test tube or in the greenhouse. The weakness of haploid plants could be due to the expression of recessive deleterious or lethal genes [32, 44]. The presence of such genes in a heterozygous state in parental genotypes could be enhanced in citrus by generalized vegetative propagation. Nevertheless, in our work, we obtained a haploid plant of clementine (Line G) manifesting much more vigorous growth than the other haploid clementine plants, and it flowered after four years. This plant has been propagated in a greenhouse on different rootstocks of citrus and its culture does not require special care. It has also been grafted in the field, where it is cultivated under the same conditions as the commercial citrus varieties.

The absence of locules has been observed very frequently in the anthers of haploid plants and the rare locules always contained abnormal pollen grains. Pummelo haploid flowers also contained fewer pollen grains than those of diploid plants and displayed very low viability [58] and similar observations have also been made in Prunus [59].

The ovaries of some of the haploid plants characterized here were abnormal, having carpellous leaves not fused with the central axis. This anatomic characteristic may prevent the pollen tube from reaching the ovule and, thus, contribute to the lower number of ovules per ovary in the haploid plant compared to the diploid plant. Thus, in addition to the abnormal meiosis associated with haploidy, the anatomical characteristics of haploid plants can explain unsuccessful hybridisations, so far, using the pollen of different citrus genotypes. Yahata et al [58] suggested that the use of haploid plants for improvement programs or genetic research was very complicated.

Use of the haploid clementine line G for further genetic studies

The development of a double haploid line of the G haploid plant was considered an important objective, within the perspective of potential use in further genetic studies. Spontaneously a double haploid plant was obtained from the haploid plant G by somatic duplication of chromosome number. Duplication of the chromosome stock of nucellar cell-producing embryos is frequent in apomictic diploid plants of citrus [60]. However, it is rare in somatic vegetative tissue of diploid plants [61] although Yamamoto and Tominaga [62] obtained a haploid-diploid periclinal chimera from a haploid plant recovered in a cross between clementine and a triploid hybrid. This double haploid line is still young and, therefore has not been characterized yet from the reproductive point of view. Haploid and double haploid plants, together with the original diploid plant and autotetraploid plants obtained in our laboratory [63] are excellent material to carry out pioneering experiments in citrus in terms of examining the genetic and epigenetic changes that can take place as a result of genic dose and for further genetic studies in relation to the whole citrus genome sequencing project.

Plant material

'Clemenules' clementine was used as female parent and 'Fortune'mandarin was selected as male parent due to its high pollen fertility and compatibility with the clementine group. Flowers of 'Fortune' mandarin was collected in preanthesis and were irradiated with gamma rays with a single dose of 500 Gy from a cobalt 60 source.

Pollination and embryo rescue

Pollinations were carried out in trees growing in a greenhouse. Three-hundred and fifty flowers of 'Clemenules' clementine were pollinated manually by placing an irradiated anther in the stigma of each flower. Fruits were collected at maturity and seeds were extracted and surface sterilized with a sodium hipochloride solution (0.5% active chlorine).

Embryos were isolated from underdeveloped seeds in aseptic conditions with the aid of a stereoscopic microscope and cultivated on Petri dishes containing the Murashige and Skoog [64] culture media with 50 g/L sucrose, 500 mg/L malt extract supplemented with vitamins (100 mg/L i-inositol, 1 mg/L pyridoxine hydrochloride, 1 mg/L nicotinic acid, 0.2 mg/L thiamine hydrochloride, 4 mg/L glycine) and 8 g/L Bacto agar (culture media MS). After germination plants were transferred to 25 × 150 mm test tubes with MS culture media without malt extract.

Embryogenic calli were cultivated in MS culture media with 40 g/L of sucrose and 1.8 g/L gelrite. Embryos obtained from these calli were cultured in MS culture media with 0.02 mg/L α-naftalenacatic acid and 1 mg/L gibberellic acid to promote germination. Cultures were maintained at 24 ± 1°C, 60% humidity and 16 h daily exposure to 40 μE m^-2 s^-1 illumination.

Plant regeneration

The germinated embryos did not develop roots or, at most, very small and weak ones. For this reason in vitro shoot tip grafting was used for plant regeneration [46]. Micrografted plants were cultivated in a liquid nutrient culture medium composed of the plant cell salt solution of Murashige and Skoog [64] with vitamins (100 mg/L i-inositol, 1 mg/L pyridoxine hydrochloride, 1 mg/L nicotinic acid and 0.2 mg/L thiamine hydrochloride) and 75 g/L sucrose. The medium was distributed into 25 × 150 mm test tubes in 25 mL aliquots. A folded paper platform, perforated at its centre for insertion of the root portion of the rootstock, was placed in the nutrient solution. The cultures were kept at a constant 24 ± 1°C and exposed 16 h daily to 40 μE m^-2 s^-1 illumination.

Ploidy level analysis

Ploidy level was determined by flow cytometry. Each sample consisted of a small leaf piece (~0.5 mm²) collected from each plant with a similar leaf piece from a diploid or triploid control plant. For embryogenic calli, two sample types were taken: one sample was just composed of a fragment of the callus, and the other from a fragment of callus with a piece of leaf from a control plant. Samples were chopped together using a razor blade in the presence of a nuclei isolation solution (High Resolution DNA Kit Type P, solution A; Partec, Münster, Germany). Nuclei were filtered through a 30-μm nylon filter and stained with a DAPI (4-6-diamine-2-phenylindol) (High Resolution DNA Kit Type P, solution B; Partec) solution. Following a 5 min incubation, stained samples were run in a Ploidy Analyzer (Partec, PA) flow cytometer equipped with a HBO 100-W high-pressure mercury bulb and both KG1 and BG38 filter sets. Histograms were analyzed using the dpac v2.0 software (Partec), which determines peak position, coefficient of variation (CV), and the relative ploidy index of the samples.

Genetic analysis

Genetic analysis was carried out with SSRs markers. The extraction of genomic DNA was done according to Dellaporta and Hicks [65] with small modifications. The plants obtained were analysed with heterozygotic markers for clementine [19, 66, 67]. These markers are positioned broadly throughout the genetic map of clementine [24].

PCR product conditions and separation were by means of vertical denaturalized electrophoresis (bis-acrylamide acrylamide 6%, urea 7 M) buffer TBE 0.5× (Tris, boric acid and EDTA 0.5 M, pH 8) in a DCodeTM Biorad, according to the methodology described by Froelicher et al. [45]. The amplified fragments were detected by means of silver staining [68].

Chromosome counts

Chromosome counts were determined according to the methodology described by D'Hont et al [69] and observations were carried out with ultraviolet light in an E800 eclipse Nikon microscope.

Morphological characterization

Thirty adult leaves located in the intermediate zone of spring shoots were taken from each plant. Area, width and foliar length were measured with LiCor 3100 C equipment.

Histological characterization

Twenty flowers of diploid 'Clemenules' clementine and the haploid plant G, were collected in preanthesis. Ovaries, styles and stigmas were fixed in FAA (formaldehyde, glacial acetic acid and alcohol 50%). The samples were embedded in paraffin, cut in portions of 10 μm and dyed with safranin and fast green according to the general methodology described by Jensen [70]. Twenty sections of the intermediate zone of each organ were analysed. Observations were made with an E800 Eclipse Nikon microscope.

Statistical analysis

A hierarchized ANOVA was carried out for each of the measured characters, applying the transformation square root (v) for the variable area (cm²). The model used was: x_ijk = μ + P_i + (G)_j(i) + ω k(ij). Later the adjusted measurements were compared using Bonferroni correction.

The normality of the variables was verified with the Kolmogorov test and an ANOVA was made of the all variables analysed between anthers, ovaries, styles and stigmas. The model used was: x_ij = μ + G_j + ω_ij.