Occurrence of viable unreduced pollen in a Begoniacollection

Angelo Dewitte Æ Tom Eeckhaut ÆJohan Van Huylenbroeck Æ Erik Van Bockstaele

Received: 29 August 2007 / Accepted: 14 January 2009 / Published online: 5 February 2009

� Springer Science+Business Media B.V. 2009

Abstract Polyploidy is widespread in plants and

has played a major role in the evolution and

diversification of the plant kingdom. Unreduced

(2n) gametes are an interesting tool for polyploidi-

sation and the creation of genetic variation in plant

breeding. Especially in ornamentals, polyploidisation

can broaden attractive features within a species. A

Begonia collection was screened on the occurrence of

2n pollen with the aid of four different techniques:

pollen size measurement, flow cytometric analysis of

nuclei isolated from germinated and non germinated

pollen, investigation of the microsporogenesis and

analysis of progeny. In ten of the 70 screened

genotypes (B. dregei, B. pearcei, B. ‘Anna Christina’,

B. ‘Bubbles’, B. ‘Florence Rita’, B. ‘Orococo’,

B. ‘Rubaiyat’, B. ‘Spatflacier’, B. ‘Tamo’ and B276),

large pollen were observed with a rather spherical than

normal ellipsoidal shape. Flow cytometric data proved

that these aberrantly shaped pollen were associated

with 2n ploidy levels, although they were not always

viable. Meiotic aberrations in these large pollen

producers resulted mainly in dyads although also

monads, triads and polyads were observed. Successful

crosses were obtained with B. dregei, B. ‘Orococo’,

and B276 as pollinators; DNA content had increased in

all or a part of the progeny. The results show that the

occurrence of 2n pollen is not a rare phenomenon in

Begonia.

Keywords 2n pollen � Flow cytometry �Germination � Microsporogenesis �Nuclei � Progeny

Introduction

Polyploids are plants with somatic cells that contain

more than twice the basic set of chromosomes.

Polyploidisation is widespread in plants and has

played a major role in the evolution and diversifica-

tion of the plant kingdom and is regarded as an

important mechanism of speciation and adaptation in

plants. Polyploids differ from their diploid seedlings

in morphological, ecological, physiological, and

cytological characteristics. These genotypic and phe-

notypic differences are caused mainly by the

increased cell size, gene dosage effect and allelic

diversity (Ramsey and Schemske 1998). Especially in

cultivated plants polyploids are present and some of

the important crops such as wheat, potato, cotton, oat,

sugarcane, banana, groundnut, tobacco, and numerous

horticultural crops are all polyploids (Ramanna and

A. Dewitte � T. Eeckhaut � J. Van Huylenbroeck (&) �E. Van Bockstaele

ILVO, Plant Unit, Applied Genetics and Breeding,

Caritasstraat 21, 9090 Melle, Belgium

e-mail: Johan.Vanhuylenbroeck@ilvo.vlaanderen.be

E. Van Bockstaele

Department for Plant Production, Ghent University,

Coupure Links 653, 9000 Ghent, Belgium

123

Euphytica (2009) 168:81–94

DOI 10.1007/s10681-009-9891-x

Jacobsen 2003). Gametes with somatic chromosome

numbers, better known as unreduced (2n) gametes, are

considered to be the driving force behind the forma-

tion of polyploids. The production of 2n gametes has

been documented in a wide variety of angiosperm

species as Dactylis glomerata, Medicago spp., Lolium

perenne, Solanum spp. or Trifolium spp. (Harlan and

de Wet 1975; Veilleux 1985; Bretagnolle and Thomp-

son 1995; Ramanna and Jacobsen 2003).

In plant breeding, the traditional chromosome

duplication is performed through the use of harmful

mitosis inhibitors like colchicine, oryzalin and triflur-

aline (Vaughn and Lehnen 1991; Hancock 1997). The

major drawback of this technique is that it largely

disables the recombination between chromosomes of

different parents in allopolyploids (Lim et al. 2000;

Ramanna and Jacobsen 2003). As a consequence, this

form of polyploidisation offers limited possibilities to

create genetic variability. Unreduced gametes can be

an interesting alternative for polyploidisation in plant

breeding. Polyploids derived from 2n gametes have

frequently shown to be more useful than chemically

induced polyploids for crop improvement (Ramanna

and Jacobsen 2003). In Lilium breeding, 2n gametes

have been successfully used for the production of

sexual polyploids from sterile Oriental 9 Asiatic

hybrids. In these hybrids, only a few hybrids produced

viable 2n pollen at variable frequencies, but they were

successfully used for the production of backcross

progeny (Barba-Gonzalez et al. 2004). In Begonia,

semperflorens begonias originate from the cross

between B. cucullata var. hookeri (syn.: B. semper-

florens) and B. schmidtiana. The diploid hybrid is

nearly sterile. With the use of unreduced gametes,

allo-tetraploids have originated and the resulting

cultivars surpassed the diploids (Horn 2004). Other

examples concerning the use of 2n gametes for crop

improvement in plant breeding have been demon-

strated among others in potato, alfalfa, red clover, and

blueberry (Ramanna and Jacobsen 2003).

Although most angiosperms produce 2n gametes,

their detection is not evident. Unreduced pollen can

be detected in four different ways (Bretagnolle and

Thompson 1995). First, unreduced pollen are gener-

ally associated with abnormally large pollen size.

This association is due to the positive correlation

between DNA content and cell volume which, in turn,

influences pollen diameter. Second, 2n pollen can be

screened flow cytometrically. With this method,

direct quantification of nuclear pollen DNA is

possible. One pollengrain contains two cells: one

vegetative and one generative cell, each containing a

vegetative and a generative nucleus (binucleate

pollen). During further pollen maturation, the gener-

ative nucleus divides into two sperm cells resulting in

a trinucleate cell. The time of division can vary

between different plant families (Russell 1991). Flow

cytometry has proven that the generative nucleus of

binucleate cells may progress to the G2 phase of the

cell cycle without dividing into two sperm nuclei

(Van Tuyl et al. 1989; Bino et al. 1990). Hence, this

nucleus obtains a double DNA content compared to

the vegetative nucleus. Third, ploidy analysis of the

progeny can reveal the presence of unreduced

gametes in parent plants. Fourth, techniques to stain

fixed anthers can reveal abnormalities of the micro-

sporogenesis which are related to unreduced gametes.

In a normal meiosis, one mother cell divides in four

daughter cells (tetrad formation). Abnormalities

results in the formation of triads, dyads or monads,

depending on the mechanism of 2n gamete formation.

Of these four techniques, only the latter two methods

can be used to detect 2n egg formation (Jongedijk

1987; Lamote et al. 2002; Estrada-Luna et al. 2004).

The genus Begonia contains more than 1,400

species with very different phenotypical characteris-

tics and chromosome numbers (Doorenbos et al.

1998). Begonia has a relative short reproduction time

and can flower throughout the year. Therefore,

Begonia would be an ideal model crop to investigate

the possible role of unreduced gametes to create

genetic variation in ornamental plant breeding. For

practical applications in plant breeding, these unre-

duced pollen must also be able to germinate and

fertilise egg cells. However, a fast and reliable

screening method to detect unreduced gametes is

needed since nothing is currently known about the

occurrence of unreduced gametes in Begonia. This

study was focussed on the detection of unreduced

pollen in a Begonia collection.

Materials and methods

Plant material

Seventy different species and accessions from a

Begonia collection at the ILVO were used in this

82 Euphytica (2009) 168:81–94

123

study (Table 1). Genotypes were grown in green-

houses, under standard conditions during winter

[20 ± 2�C, extended natural photosynthetic period

to 16 h day-1 (Photosynthetic photon flux density

30 lmol m-2 s-1)]. Crossing hybrids were labelled

with a ‘B’ number (as B201).

Pollen size measurements and germination ratio

Opened flowers of Begonia plants were collected and

pollen anthesis was stimulated by overnight illumi-

nation. Subsequently, anthers of at least five flowers

were submersed in pollen germination medium (10%

sucrose, 0.01% H3BO3, 0.01% CaCl2, 0.02%

MgSO4 � 7H2O, 0.01% KH2PO4, 0.01% chloram-

phenicol); the solution was vortexed and filtered

through a nylon filter of 50 lm mesh size. A drop of

the pollen filtrate was pipetted on a slide, covered

with a cover slip and examined under a light

microscope (Leica DMIRB). The longest axis of the

ellipsoidal pollen was measured digitally using

software Image manager IM 500 (Leica). At least

30 pollen of each genotype were measured to

determine the typical pollen size range. When a

subpopulation of large pollen was observed, its

percentage was estimated based on at least 300

pollen. The germination ratio was determined micro-

scopically after overnight germination.

Begonia genotypes producing large pollen were

screened at least three times at different time intervals

(1, 2, and 4 weeks after the first observation) to

monitor the presence of the large pollen fraction over

time.

Flow cytometry

Flow cytometry was used to compare the relative DNA

content of somatic leaf tissue and pollen of a specific

Begonia genotype. This technique allows rapid deter-

mination of the relative DNA content of individual

nuclei by measuring the fluorescence of a fluoro-

chrome [40,6-diamidino-2-phenylindole (DAPI)] that

binds to DNA (Galbraith et al. 1983).

The release of individual nuclei of somatic leaf

tissue was done according to De Schepper et al.

(2001). Leaf peaces of 20 mm2 are cut from the

youngest leaves of Begonia and chopped in 400 ll of

solution A (0.1M citric acid and 0.5% Tween 20)

using a sharp razor blade. After filtration through a

nylon filter of 50 lm mesh size, 800 ll of solution B

(0.4M Na2HPO4 and 1 lM DAPI) is added for

staining. Nuclei were analysed using flow cytometry

(Partec PAS III, Munster, Germany).

Two methods are used to isolate nuclei from

pollen. The amount of collected anthers varied from

genotype to genotype, dependent on the amount of

flowers present and the amount of pollen produced by

the genotype. Usually, anthers from 20 to 50 flowers

were collected. In the first method, collected anthers

are submersed in 5 ml nuclei isolation buffer (NIB)

described by Pan et al. (2004), and subsequently

vortexed and filtered through a nylon filter of 50 lm

mesh size. The pollen filtrate was treated 10 min with

an ultrasonic device (Biologics Inc., ultrasonic

homogenizer, model 150 V/T), output 40, using

80% pulses. Every 2.5 min, a 15 s pause was

incorporated. The treated solution was centrifuged

at 500g for 20 s, and the supernatant was filtered

through a 11 lm nylon filter (Millipore, Bedford,

Mass., USA) and a 8 lm SPI-poreTM filter (Structure

probe Inc., West Chester, USA). The filtrate was

centrifuged 4 min at 2,000g and the pellet was

resuspended in 400 ll solution A containing 0.5%

Triton X. The solution was carefully shaken for

15 min and centrifuged 4 min at 700g. The pellet was

resuspended in 400 ll solution A and 800 ll solution

B and analysed with flow cytometry.

In the second method, anthers are submersed in

germination medium, vortexed and filtered through a

nylon filter of 50 lm mesh size. After overnight

germination, the solution was centrifuged (6 min at

6,800g), and the pellet resuspended in a hypotonic

buffer (0.01M Tris pH 7.5, 0.01M NaCl, and

0.0045M MgCl2) according to Lafountain and Mas-

carenhas (1972) for 30 min. The treated solution was

centrifuged at 500g for 20 s; the supernatant filtered

and furthermore treated as in the first method. The

obtained sample was analysed using flow cytometry.

Using the two different nuclei isolation methods,

two equally sized peaks were obtained in the flow

cytometric histogram. The first peak (1C) was gener-

ated by a population of nuclei containing half of the

DNA amount of somatic leaf nuclei, while nuclei

associated with the second peak (2C) contained the

same DNA amount of nuclei derived from somatic leaf

tissue. They respectively correspond to a population of

normal vegetative nuclei (VR) and a population of

normal generative nuclei (GR) in the G2-phase. At this

Euphytica (2009) 168:81–94 83

123

Table 1 Begonia accessions used throughout this study: origin and pollen characteristics

Genotype Origin Normal pollen

size (lm)

Abnormal

pollen (%)

Abnormal pollen

size (lm)

Germination

(%)

2n nuclei

(%)

B. albo-picta IL 18–22 0 21 0 (A)

B. boliviensis PD 17–21 0 57 0 (B)

B. bowerae USDA –y – – –

B. cucullata PD 16–20 0 81 0 (A)

B. coccinea IL 18–22 0 47 0 (A)

B. corallina IL 21–25 0 36 0 (A)

B. diadema IL 18–23 0 55 0 (A)

B. dregei IL 23–27 6z,x 27–30 22 –

B. echinosepala var. elongatifolia JR 16–21 0 93 0 (A)

B. fischeri USDA 23–28 0 78 0 (B)

B. heracleifolia IL –y – – –

B. listada IL –y – – –

B. maculata JR 19–23 0 17 0 (B)

B. manicata JR 18–22 0 81 0 (A)

B. odorata IL 18–23 0 23 0 (B)

B. partita JR 21–25 0 5 0 (A)

B. pearcei BE 19–23 8z 25–29 81 9 (B), 0 (A)

B. undulata JR 19–23 0 8 0(A)

B. solananthera IL 15–19 0 87 0 (A)

B. soli-mutata JR 15–19 0 37 0 (A)

B. subvillosa var. leptotricha IL 17–20 0 95 0 (A)

B. tomentosa JR 18–22 0 56 0 (B)

B. ulmifolia IL 15–19 0 21 0 (A)

B. venosa IL 16–21 0 84 0 (A)

B. ‘Anna Christina’ LG 20–25 2z 25–28 16 0 (A)

B. ‘Art Hodes’ LG 17–20 0 6 0 (A)

B. ‘Barbara Ann’ LG 22–26 0 0 0 (B)

B. ‘Beauregard’ IL 20–24 0 24 0 (B)

B. ‘Bubbles’ LG 23–29 14z 30–33 5 0 (A), 12 (B)

B. ‘Cracklin Rosie’ LG 20–24 0 19 0 (B)

B. ‘Dark Rex’ IL No pollen 0 0 0

B. ‘Ebony’ IL 23–28 0 43 0 (A)

B. ‘Eunice Gray’ IL Bad pollen 0 0 0

B. ‘Florence Rita’ LG 22–34 0z 0 100 (B)

B. ‘Ginny’ JR Bad pollen 0 0 0

B. ‘Honeysuckle’ LG 19–24 0 82 0 (A,B)

B. ‘Maxine Wilson’ IL 22–26 0 9 0 (A)

B. ‘Orange Rubra’ JR 20–24 0 92 0 (A)

B. ‘Orococo’ JR 14–18 88z 20–26 9 43 (A)

B. ‘Ramke’ IL 22–27 0 19 0 (A)

B. ‘Richardsiana’ LG 21–27 0 54 0 (A)

B. ‘Rubaiyat’ JR 22–27 1z 28–30 9 0 (A)

B. ‘Spatflacier’ IL 21–25 27z 26–29 49 26 (A)

B. ‘Tamo’ JR 22–28 0z 1 19 (A)

84 Euphytica (2009) 168:81–94

123

stage, DNA has been replicated but no cytokinesis

took place yet. As a consequence, vegetative nuclei of

unreduced pollen (VU) could be detected at the 2C

level (as GR) while generative nuclei of unreduced

pollen (GU) could be detected at the 4C level. With

VU theoretically equal to GU, the percentage 2n

pollen can be calculated according to the following

taxa GU

GRþGU¼ GU

GRþVU¼ 4Cnuclei

2Cnuclei� 100: For the calcula-

tion, generative nuclei are preferred above vegetative

nuclei because of the lower amount of debris present at

4C and 2C compared to 1C.

Microsporogenesis

Immature anthers were removed from young floral

buds and excised on a slide in a drop of 1 lM DAPI in

a sodium citrate solution (0.3M NaCl, 30 lM sodium

citrate, pH 7.0 in milliQ-water). After removal of the

anthers, a coverslip was added and the preparation

was observed through light or fluorescence micros-

copy. B. fisheri, B243, and B. ‘Orange Rubra’ were

used as negative controls, as they don’t produce

unreduced pollen. At least 150 cells were scored.

Table 1 continued

Genotype Origin Normal pollen

size (lm)

Abnormal

pollen (%)

Abnormal pollen

size (lm)

Germination

(%)

2n nuclei

(%)

B201 IL Bad pollen 0 0 0

B202 IL Bad pollen 0 0 0

B208 IL Bad pollen 0 0 0

B210 IL 18–22 0 11 0 (B)

B211 IL 19–23 0 12 0 (A)

B212 IL 21–28 0 29 0 (A)

B215 IL No pollen 0 0 0

B221 IL 22–28 0 2 0 (A)

B230 IL 22–25 0 34 0 (A)

B232 IL 21–26 0 5 0 (A)

B238 IL 22–27 0 79 0 (A)

B239 IL 21–26 0 9 0 (A)

B243 IL 19–22 0 74 0 (A)

B246 IL 20–26 0 2 0 (A)

B247 IL 20–26 0 5 0 (A)

B251 IL Bad pollen 0 0 0

B255 IL Bad pollen 0 0 0

B258 IL 19–22 0 75 0 (A,B)

B259 IL 21–24 0 19 0 (A,B)

B268 IL 19–22 0 11 0 (A)

B273 IL 20–23 0 24 0 (A)

B276 IL 26–32 0z 52 100 (A)

B277 IL 15–19 0 87 0 (A)

B278 IL 16–20 0 5 0 (A)

B280 IL 16–20 0 92 0 (A)

B283 IL 16–19 0 56 0 (A)

A Flow cytometrical detection after an osmotic shock on germinated pollen; B flow cytometrical detection after treatment of pollen

with a sonicator. BE Benary (Ernst benary Samenzucht GmbH); IL existing ILVO collection; LG Logee’s greenhouse; PD plant

delights nursery; JR le jardin de rochevieille; USDA united states department of agriculturex Temporary detection; no flow cytometric data obtainedy Male flower drops before opening; no pollen could be investigatedz Spherical pollen

Euphytica (2009) 168:81–94 85

123

Progeny analysis

Crosses and selfings were performed with B. dregei,

B. pearcei, B. ‘Orococo’, B. ‘Tamo’, B. ‘Bubbles’,

B. ‘Florence Rita,’ B276, B. ‘Spatflacier’ and the

control B243. Five flowers per genotype were

pollinated. Obtained seeds were sterilized with 10%

NaOCl and 0.05% teepol for 20 min, rinsed three

times with autoclaved water and initiated on 1/3 MS

(Murashige and Skoog 1962) medium supplemented

with 20 g sucrose, 100 mg Meso-inositol, 1 mg

nicotine, 1 mg glycine, 1 mg GA3, 0.5 mg thiamine,

0.5 mg pyridoxine, and 6 g agarose (pH 6.2).

Cultures were maintained at 23 ± 2�C under a 16 h

photoperiod (40 lmol m-2 s-1 photosynthetic active

radiation) and finally acclimatized in the greenhouse.

Leaves of each seedling were analysed through flow

cytometry and compared to the leaf DNA content of

both seed and pollen parents. A histogram was made

of the pollen parent as an external reference to

calibrate the flow cytometer. Seedlings originating

from normal gametes have a nuclear DNA content

intermediate to those of the parent plants. On the

other hand, a seedling created from a 2n and a normal

gamete exhibits a DNA content which is the sum of

the DNA content of the 2n gamete producing plant

and half of the DNA amount of the partner.

Results

Pollen size measurements and germination

Pollen size largely depends on the Begonia genotype

(Table 1). In most genotypes, there was a single pollen

population with a difference of 4–6 lm between

smallest and largest pollen. In case of B. ‘Orococo’,

B. ‘Bubbles’, B. ‘Anna Christina’, B. pearcei,

B. dregei, B. ‘Spatflacier’, and B. ‘Rubaiyat’, two pollen

populations with different pollen size were detected. In

these genotypes, the larger fraction comprised rather

spherical pollen (Fig. 1). In B. ‘Florence Rita’ and B276

all pollen were very large and spherical instead of

ellipsoidal. Also pollen of B. ‘Tamo’ were rather

spherical but were smaller. The frequency of these large

pollen was very genotype dependent, and varied from

1% (B. ‘Rubaiyat’) to 100% (B. ‘Florence Rita’ and

B276).

In B. dregei, large pollen have only been observed

during two consecutive weeks. In the other afore-

mentioned genotypes, the presence of large or

spherical pollen has been confirmed repeatedly.

Nine of the 70 genotypes (B. ‘Dark Rex’, B. ‘Eunice

gray’, B. ‘Ginny’, B201, B202, B208, B215, B251,

B255) produced no or shrunken pollen. Three geno-

types (B. heracleifolia, B. bowerae and B. listada)

produced male flowers that dropped before opening.

Pollen germination in the other genotypes was very

variable and varied from 1% in B. ‘Tamo’ to 95% in

B. subvillosa var. leptotricha. Only pollen of

B. ‘Florence Rita’ were unable to germinate.

Flow cytometric analysis of pollen

Using two different nuclei isolation methods, each

time two equally sized peaks were obtained, although

flow cytometric histograms contained more debris

after the use of the sonicator (Fig. 2). They respec-

tively correspond to a population of normal

vegetative nuclei and a population of normal gener-

ative nuclei. Indeed, the presence of two nuclei was

observed in pollen and pollen germ tube, both with a

different fluorescence intensity (Fig. 3). For B. pear-

cei, B. ‘Florence Rita’ and B210, the first population

was reduced in size or even totally absent compared

to the second one after use of the sonicator.

When nuclei were isolated from germinated pol-

len, a peak at the 4C level could be observed in

B. ‘Orococo’, B. ‘Tamo’, B. ‘Spatflacier’, and B276

(Table 1; Fig. 2) with a ratio 4C/2C nuclei of 0.19 in

Fig. 1 Pollen of B. ‘Bubbles’ with large spherical pollen,

normal ellipsoidal pollen and bad, shrunken pollen

86 Euphytica (2009) 168:81–94

123

B. ‘Tamo’, 0.43 in B. ‘Orococo’, 0.26 in B. ‘Spat-

flacier’ and 1 in B276. In other large pollen producers

(LPP) as B. ‘Florence Rita’, B. ‘Bubbles’ or

B. pearcei, the presence of these nuclei could only

be detected after use of the sonicator, indicating that

these large pollen do not germinate. No signal at the

4C level was obtained when the genotype did not

produce large or spherical pollen. In B. dregei, no

data were obtained, as large pollen were not consis-

tently produced in this genotype.

Microsporogenesis

During normal microsporogenesis, cytokinesis and

cell wall formation started only after telophase II,

meaning after the division of the four nuclei

(‘simultaneous type’ of meiosis). This resulted in

four microspores (tetrad) each containing one nucleus

(Fig. 4).

During microsporogenesis of all LPP, variable

frequencies of monads, triads and mainly dyads were

observed (Table 2; Fig. 4), increasing along with the

amount of large pollen present. In B. ‘Tamo’, however,

the size of the microspores was very variable (even in

tetrads) and several pentads, hexads or abnormal triads

Fig. 2 Flow cytometric

analysis of leaf and pollen

nuclei. On the X-axis, the

channel numbers of the flow

cytometric distribution

analyser are plotted, the

Y-axis represents the

number of DAPI

fluorescence signals

recorded per channel. aSomatic leaf tissue of the

genotype B258. b Normal

(reduced) pollen nuclei

released after osmotic shock

of germinated pollen

showing the peaks

according with the

vegetative (V) and the

generative (G) nuclei

populations of the same

genotype. c Normal

(reduced) pollen nuclei

released after ultrasonic

treatment. d Begoniagenotype B276 producing

only 2n gametes. The nuclei

were released after an

osmotic shock on

germinated pollen

Fig. 3 Two nuclei with different fluorescence intensity are

visualised within pollen. The more intensely stained particle is

the generative nucleus, the lesser fluorescing is the vegetative

nucleus

Euphytica (2009) 168:81–94 87

123

(containing two large and one small microspore) were

seen (Fig. 4). Only in B. dregei, no dyads but triads

were observed next to tetrads. Again, this was only

temporary, as in later observations, only tetrads (or

rarely pentads) were observed.

Formation of triads was already visible before

cytokinesis, where three nuclei were present in a

pollen mother cell (PMC). After cytokinesis, this

resulted in two smaller and one bigger microspore.

Dyads usually contained one nucleus in each micro-

spore (Type I dyad). However, two nuclei per

microspore were observed in B276 (Type II dyad).

Also monads could be divided in two types contain-

ing one (Type I) or more (Type II) nuclei. Monads

were mainly observed in B. ‘Florence Rita’. Flow

cytometric analysis (sonicator) of B. ‘Florence Rita’

pollen showed indeed a very small accumulation of

nuclei at the 8C level, with a ratio 8C/4C of 0.03.

Progeny analysis

Crosses resulting in offspring are described in

Table 3. All seedlings of the control cross

B243 9 B243 exhibited no increased nuclear DNA

content, as expected. When B243 was crossed with

B. ‘Maxine Wilson’ (which has a twice as large DNA

content), all seedlings exhibited an intermediate DNA

content of about 1.59 the DNA content of B243,

which can be expected from normal gametes.

Analysis of B. soli-mutata 9 B. ‘Orococo’, B. ‘Art

Hodes’ 9 B. ‘Orococo’, B. corallina 9 B. dregei and

B. ‘Bubbles’ 9 B276 showed an increased DNA

content in all (B. ‘Bubbles’ 9 B276), a majority

(B. soli-mutata 9 B. ‘Orococo’, B. ‘Art Hodes’ 9

B. ‘Orococo’) or a part (B. corallina 9 B. dregei) of

the progeny, while the DNA content of the parent

plants was very similar. The DNA content of these

seedlings was about 1.5 higher compared to the

parent plants, what could be expected from a cross

between a normal and a 2n gamete.

Comparison between the four methods

Table 4 summarizes the data obtained with the

different detection methods for the 2n pollen pro-

ducers and controls. The presence or absence of

Fig. 4 Microsporogenesis in Begonia. a Normal tetrad. b Triad with one large and two small microspores. c Dyad. d Abnormal triad

in B. ‘Tamo’ consisting of two large microspores and one small microspore

88 Euphytica (2009) 168:81–94

123

unreduced gametes was in all genotypes confirmed by

the use of the different methods.

In none of the control genotypes evidence was

found for 2n gamete production. Within the 2n pollen

producers, the frequency of 2n pollen depended on

the used method. B. pearcei for instance, shows

frequencies between 3.7 and 9%. However, no 2n

pollen were detected using the flow cytometric

method on germinated pollen. Similar results were

obtained for B. ‘Anna Christina’, B. ‘Bubbles’, and

B. ‘Rubaiyat’. In B. ‘Florence Rita’ and B276, only

2n pollen were detected based on pollen shape and

flow cytometric data. Nevertheless, during the micro-

sporogenesis also a low number of tetrads (which

should result in normal pollen) were formed in these

two genotypes. In B. ‘Orococo’, 96% of the seedlings

resulted from unreduced pollen, while only 43% of

the viable pollen are unreduced. Remarkable results

were obtained with B. ‘Tamo’, where all pollen were

spherical but only a part of these pollen was

unreduced. Based on microsporogenesis, this fre-

quency even does not exceed 4%.

Discussion

After studying a collection of 70 Begonia genotypes,

unreduced pollen have been detected in ten different

accessions based on pollen size and shape, flow

cytometry, microsporogenesis and progeny analysis.

Table 3 Progeny analysis

from successful crosses

with large pollen producers,

and B243 as control

Cross Number of seedlings

evaluated

Number of seedlings

obtained from 2n gametes$ #

B243 B243 30 0

B. ‘Maxine Wilson’ B243 50 0

B. soli-mutata B. ‘Orococo’ 212 203

B. ‘Art Hodes’ B. ‘Orococo’ 64 63

B. corallina B. dregei 3 1

B. ‘Bubbles’ B276 37 37

Table 2 Meiosis aberrations (%) in different Begonia genotypes

Tetrad Triad Dyada Monada Othersb

Type I Type II Type I Type II

Controls

B. fischeri 99 0 0 0 0 0 1

B. ‘Orange Rubra’ 99.5 0 0 0 0.5 0 0

B243 100 0 0 0 0 0 0

2n pollen producers

B. dregei 89.5 10 0 0 0 0 0.5

B. pearcei 89 3 5 0 1 0 2

B. ‘Anna Christina’ 94 0.5 2 0 1 0 2.5

B. ‘Bubbles’ 81 4 12.5 0 1 1 0.5

B. ‘Florence Rita’ 5 1 82 0 3 9 0

B. ‘Orococo’ 42 4 52 0 0 0 2

B. ‘Rubaiyat’ 97.5 0 1.5 0 0.5 0 0.5

B. ‘Spatflacier’ 82 2 10 1 1 0 4

B. ‘Tamo’ 55 1 4 0 1 0 39

B276 4 2 87 2 2 4 0

a Type I one nucleus per microspore; type II more than one nucleus per microsporeb Pentads, hexads or abnormal triads with two large and one small microspore

Euphytica (2009) 168:81–94 89

123

These results show that unreduced pollen are not

uncommon in Begonia, as a relatively high number of

investigated plants (14%) produced unreduced pollen.

Since chromosome numbers in Begonia are very

variable, polyploidisation must have played a major

role in the diversification of the genus. This was also

concluded by Oginuma and Peng (2002), who showed

the presence of chromosomes with secondary, terti-

airy, and/or small constrictions of polyploids within

Taiwanese Begonia. Some of the examined genotypes

(B. ‘Orococo’, B. ‘Tamo’) are interspecific hybrids

(Tebbitt 2005). It is well known that interspecific

hybrids frequently produce unreduced gametes in

other plant genera (Ramanna and Jacobsen 2003).

Comparison between the different methods used in

this study showed that the frequency of 2n pollen

depended on the used method. However, it must be

mentioned that frequencies may vary in a certain

genotype during time due to environmental fluctua-

tions (Bretagnolle and Thompson 1995; Crespel et al.

2006).

Microscopical determination of pollen grain sizes

is a very fast method for the detection of unreduced

pollen. Pollen of B. dregei, B. pearcei, B. ‘Orococo’,

B. ‘Rubaiyat’, B. ‘Anna Christina’, B. ‘Spatflacier’,

and B. ‘Bubbles’ could be clearly subdivided in two

populations with different pollen size. It is well

known that 2n pollen grains are generally larger than

normal grains. The presence of giant grains fre-

quently is an indication for the production of 2n

pollen (Sala et al. 1989; Van Tuyl et al. 1989;

Bretagnolle 2001; Becerra Lopez-Lavalle and Orjeda

2002) and our results proved that this is also the case

in Begonia. However, a larger pollen diameter is no

unequivocal proof for a doubled DNA content.

Moreover, supplementary evaluation of pollen via-

bility is necessary. Another disadvantage is the broad

overlap in size distribution between small and large

pollen in some genera like grasses. In these cases, the

frequency of unreduced pollen based on pollen size is

difficult to determine (Jansen and Den Nijs 1993).

Our results show that besides size also pollen shape

(spherical in stead of ellipsoidal) was associated with

the ploidy level, which facilicitates the determination

of 2n pollen. These spherical pollen had a larger

diameter. Akutsu et al. (2007) observed in lilies that

Table 4 Comparison between the different methods used for frequency determination of 2n pollen

Microscopical evaluation Flow cytometry Meiosisa Progeny analysisb

Size Shape Osmotically Sonicator

Controls

B. fischeri 0 0 – 0 0 –

B. ‘Orange Rubra’ 0 0 0 – 0 –

B243 0 0 0 – 0 0

2n pollen producers

B. dregei 6 6 – – 3 33

B. pearcei 8 8 0 9 4 –

B. ‘Anna Christina’ 2 2 0 – 1 –

B. ‘Bubbles’ 5 5 0 12 8 –

B. ‘Florence Rita’ 0 100 – 100 89 –

B. ‘Orococo’ 88 88 43 – 37 96

B. ‘Rubaiyat’ 1 1 0 – 1 –

B. ‘Spatflacier’ 27 27 26 – 7 –

B. ‘Tamo’ 0 100 26 – 4 –

B276 0 100 100 – 90 100

‘–’ No data availablea The frequency was calculated based on the numbers of tetrads, triads, dyads, and monads observedb Proportion between seedlings obtained from unreduced gametes and total number of seedlings. For B. ‘Orococo’, both crosses with

B. soli-mutata and B. ‘Art Hodes’ (Table 3) were evaluated simultaneously (276 seedlings). For B. dregei, only three seedlings could

be evaluated

90 Euphytica (2009) 168:81–94

123

induced unreduced pollen had rather a spherical than

an elliptical shape. In our study, the spherical shape

of the pollen was a strong indication in B. ‘Florence

Rita’ and B276 that all these pollen were unreduced,

as was confirmed with the other methods. Only in

B. ‘Tamo’, no two populations with different pollen

size could be observed microscopically, while flow

cytometric data showed the presence of both ‘n’ and

‘2n’ nuclei. Possibly, there was a broad overlap in

size distribution between normal and large pollen,

while both populations were rather spherical. In this

case, the accuracy of using pollen size or shape to

determine the frequency 2n pollen was rather poor. In

B. dregei, large pollen were observed only temporary,

but also meiotic studies demonstrated the (temporary)

presence of triads within this genotype. It remains

unclear what causes this observational anomality,

although environmental changes (e.g., temperature)

can have an effect on 2n pollen production (Negri and

Lemmi 1998; Crespel et al. 2006).

Flow cytometric investigation of pollen DNA

content revealed the presence of two populations of

even size. The first contained the vegetative nuclei,

with half of the DNA content of leaf nuclei; the latter

was composed of generative nuclei with the same

DNA content as the leaf tissue. This indicates that the

haploid generative nucleus is in the post replication

stage of cell division (G2) and generates the 2C-peak

(van Tuyl et al. 1989; Bino et al. 1990). DAPI

staining of Begonia pollen indeed revealed a more

(generative) and a less (vegetative) intensely stained

nucleus in pollen or in the pollen tube. The generative

nucleus divides into two sperm cells, although the

time of division can vary between different plant

families (Russell 1991). In our observations, at the

moment of detection, the generative nucleus had

already duplicated its amount of chromosomes, but

had not divided yet. As a consequence, vegetative

nuclei of unreduced pollen exhibit the same DNA

content of the leaf reference and generative nuclei

from normal pollen (2n), while generative nuclei

from unreduced pollen have a doubled DNA con-

tent compared to the leaf reference (4n). In a pollen

population with normal and unreduced pollen,

generative ‘normal’ nuclei and vegetative ‘unre-

duced’ nuclei can not be distinguished from each

other.

The main bottleneck to detect unreduced pollen

with flow cytometry, is finding a suitable way to

isolate pollen nuclei. Indeed, the presence of a

complex outer exine layer on the pollen surface

makes pollen strongly resistant to several physical

and chemical agents (Bohne et al. 2003). Also in lily

and Brassica napus, unreduced pollen have been

detected using flow cytometry (Van Tuyl et al. 1989;

Pan et al. 2004), although no direct information on

their viability was obtained. For practical plant

breeding, only pollen able to germinate are important.

Flow cytometric detection of nuclei released after

osmotic shock of germinated pollen offers a powerful

screening method to prove the presence of viable

unreduced pollen. A prerequisite for this is that not

germinated pollen do not release their nuclei during

treatment, which is the case in Begonia. Indeed, no

signal was obtained when pollen of B. ‘Florence Rita’

(which do not germinate) were treated osmotically.

Although ultrasonically mediated release of nuclei

can also be used to detect unreduced pollen, it

provides no direct information on their viability.

Ultrasonic treatment also causes more debris in a flow

cytometric histogram. Nuclei that were damaged by

the treatment generate noise that (partially) covers the

peak of the vegetative nuclei. In some cases, the

vegetative nuclei peak was reduced in size or

completely absent, which suggests that this nucleus

is less stable when the sonicator protocol is used.

Decreasing the time of ultrasonic treatment decreased

not only the amount of damaged nuclei, but also the

amount of released nuclei (data not shown). By com-

bining both isolation techniques, it was shown that

unreduced pollen were not always viable, as demon-

strated in B. pearcei, B. ‘Bubbles’, and B. ‘Florence

Rita’.

Although sample preparation for flow cytometry

requires more time compared to microscopic evalu-

ation, the presence of unreduced pollen can be

unambiguously confirmed. When the nuclei are

released after sonicator treatment, the frequency

should be similar to the number of large pollen,

assuming that both n and 2n pollen are even sensitive

to disruption. On contrary, when the nuclear release

of germinated pollen is used, significant lower

frequencies might be obtained, as was observed in

B. ‘Bubbles’ and B. pearcei. Both genotypes form

unreduced gametes but they did not germinate. This

screening method can be used routineously in breed-

ing programs, since only viable unreduced pollen are

of importance.

Euphytica (2009) 168:81–94 91

123

2n pollen producers are characterised by abnor-

malities during meiosis (Islam and Shepherd 1980;

Bretagnolle and Thompson 1995; Pagliarini et al.

1999; El Mokadem et al. 2002). The identification of

the mechanisms behind the formation of unreduced

gametes is complex, because different cytological

mechanisms may operate within one individual. In

Begonia, as in most dicots, a ‘simultaneous’ type of

meiosis is active: the pollen mother cells forms four

nuclei after two meiotic divisions; subsequently

cytokinesis and cell wall formation lead to the

formation of tetrads. Each individual microspore

contains half of the DNA amount of the original

mother cell. Abnormalities during meiosis were

characterised by the accumulation of mainly dyads,

with exception of B. dregei. These dyads usually

contain one nucleus in each microspore, although

occasionally 2-nuclear dyads were observed in

B276, which can be explained by an irregular

cytokinesis.

In some rare cases, meiotic aberrations result in

the formation of monads with one (type I) or more

nuclei (type II). These ‘jumbo pollen’ contain 8n

generative nuclei as detected with the flow cytometer

in B. ‘Florence Rita’, but were not viable. In contrast,

normal tetrads were observed in B. ‘Florence Rita’

and B276, while no normal n pollen were detected

with flow cytometry. In this case, premeiotic mitosis

could have led to a 4n pollen mother cell (Bretagnolle

and Thompson 1995). Thus, despite normal meiosis,

the pollen grain could have retained the initial

somatic chromosome number. Another possibility is

that microspores in tetrads originate from an unequal

chromosome division, which usually leads to sterile

pollen (Bretagnolle and Thompson 1995). Hexads,

pentads, tetrads, and triads with abnormal microspore

size were typical results of meiotic aberrations in

B. ‘Tamo’. This can be explained by abnormal

chromosome divisions during anaphase I and II,

resulting in small micronuclei. These polyads have

been observed in several other plant genera, as Citrus

or Nicotiana (Del Bosco et al. 1999; Trojack-Goluch

and Berbec 2003). It also explains the low pollen

fertility of B. ‘Tamo’. More detailed meiotic studies

are currently performed to identify the mechanisms

of meiosis aberrations in these genotypes.

Analysis of the meiosis does not provide any

information about pollen viability. Moreover, what is

observed at the end of the meiosis, is not necessarily

reflected in the pollen. Tetrads observed in B. ‘Florence

Rita’ and B276 did not end up as normal pollen, as they

could not be detected by the other methods. Hence, the

final 2n pollen frequency is not only the result of the

proportion between tetrads, triads, diads, and monads,

but also of the balanced chromosome segregation

during the formation of these meiotic products. When

the chromosome seggregation is not balanced,

the formed pollen will become bad or shrunken

(Bretagnolle and Thompson 1995). Therefore, scor-

ing the meiotic products is not the best method for

frequency determination of the final 2n pollen

frequency, although in most investigated genotypes,

the obtained frequency was very similar to the

frequencies obtained with the other methods. Some

obvious exceptions on this were B. ‘Tamo’ and

B. ‘Spatflacier’.

Progeny analysis directly indicates the existence of

viable unreduced pollen in parent plants. Neverthe-

less, it is possible that unreduced pollen in parent

plants remain unnoticed because of preferential

pairing between normal gametes. As a drawback, it

is very time consuming and no information about the

production frequency in the parent plant can be

obtained (Bretagnolle and Thompson 1995) as there

might be a difference in pollen viability, germination

speed or pollen tube growth between n and 2n pollen.

Van Breukelen (1982) showed that in Solanum, 2x

pollen grew faster than x pollen what can strongly

increase the number of seedlings with an increased

DNA content. A clear example of this was observed

in B. ‘Orococo’, where the number of meiotic

polyploids was much higher than could be expected

from flow cytometry.

The presence of unreduced pollen is not a rare

phenomenon in Begonia, but frequencies should be

considered carefully. The occurrence of unreduced

gametes by a specific method was confirmed by at

least one other method during this study. For

breeding purposes, the most reliable results are

obtained after nuclei isolation from germinated

pollen followed by flow cytometric analysis. This

frequency represent the percentage 2n pollen able to

germinate, which may lead to the formation of

meiotic polyploids.

Acknowledgment The authors want to thank the Institute for

Agricultural and Fisheries Research (ILVO) and IWT-Flanders

(project 40696) for supplying financial support.

92 Euphytica (2009) 168:81–94

123

References

Akutsu M, Kitamura S, Toda R, Miyajima I, Okazaki K (2007)

Production of 2n pollen of Asiatic hybrid lilies by nitrous

oxide treatments. Euphytica 155:143–152. doi:10.1007/

s10681-006-9317-y

Barba-Gonzalez B, Lokker AC, Lim K-B, Ramanna MS, Van

Tuyl JM (2004) Use of 2n gametes for the production of

sexual polyploids from sterile Oriental 9 Asiatic hybrids

of lilies (Lilium). Theor Appl Genet 109:1125–1132. doi:

10.1007/s00122-004-1739-0

Becerra Lopez-Lavalle LA, Orjeda G (2002) Occurrence and

cytological mechanism of 2n pollen formation in a tetra-

ploid accession of Ipomoea batatas (sweet potato).

J Hered 93:185–192. doi:10.1093/jhered/93.3.185

Bino RJ, van Tuyl JM, de Vries JN (1990) Flow cytometric

determination of relative nuclear DNA contents in bi-

cellulate and tricellulate pollen. Ann Bot (Lond) 65:3–8

Bohne G, Richter E, Woehlecke H, Ehwald R (2003) Diffusion

barriers of tripartite sporopollenin microcapsules prepared

from pine pollen. Ann Bot (Lond) 92:289–297. doi:

10.1093/aob/mcg136

Bretagnolle F (2001) Pollen production and spontaneous

polyploidization in diploid populations of Anthoxanthumalpinum. Biol J Linn Soc Lond 72:241–247. doi:10.1111/

j.1095-8312.2001.tb01314.x

Bretagnolle F, Thompson JD (1995) Gametes with the somatic

chromosome number: mechanisms of their formation and

role in the evolution of autoploid plants. New Phytol

129:1–22. doi:10.1111/j.1469-8137.1995.tb03005.x

Crespel L, Ricci S, Gudin S (2006) The production of 2n pollen

in rose. Euphytica 151:155–164. doi:10.1007/s10681-006-

9136-1

De Schepper S, Leus L, Mertens M, Heursel J, Van Bockstaele

E, De Loose M (2001) Flow cytometric analysis of ploidy

in Rhododendron subgenus Tsutsusi. HortScience 36:125–

127

Del Bosco SF, Tusa N, Conicella C (1999) Microsporogenesis

in a Citrus interspecific tetraploid somatic hybrid and it

fusion parents. Heredity 83:373–377. doi:10.1038/sj.hdy.

6885770

Doorenbos J, Sosef M, de Wilde J (eds) (1998) The sections of

Begonia (Studies in Begoniaceae VI). Backhuys Pub-

lishers, Leiden

El Mokadem H, Crespel L, Meynet J, Gudin S (2002) The

occurrence of 2n pollen and the origin of sexual polyp-

loids in haploid roses (Rosa hybrid L.). Euphytica

125:169–177

Estrada-luna AA, Garcia-Aguilar M, Vielle-Caldaza JP (2004)

Female reproductive development and pollen tube growth in

diploid genotypes of Solanum cardiophyllum Lindl. Sex

Plant Reprod 17:117–124. doi:10.1007/s00497-004-0219-7

Galbraith DW, Harkins KR, Maddox JM, Ayres NM, Sharma

DP, Firoozabady E (1983) Rapid flow cytometric analysis

of the cell cycle in intact plants tissues. Science

220:1049–1051. doi:10.1126/science.220.4601.1049

Hancock J (1997) The colchicine story. HortScience 32:1011–1012

Harlan JR, de Wet JMJ (1975) The origin of polyploidy. Bot

Rev 41:361–390. doi:10.1007/BF02860830

Horn W (2004) The patterns of evolution and ornamental plant

breeding. Acta Hortic 651:19–31

Islam AKMR, Shepherd KW (1980) Meiotic restitution in

wheat-barley hybrids. Chromosoma 79:363–372. doi:

10.1007/BF00327326

Jansen RC, Den Nijs APM (1993) A statistical mixture model for

estimating the proportion of unreduced pollen grains in

perennial ryegrass (Lolium perenne L.) via the size of pollen

grains. Euphytica 70:205–215. doi:10.1007/BF00023761

Jongedijk E (1987) A rapid methyl salicylate clearing tech-

nique for routine phase contrast observations on female

meiosis in solanum. J Microsc 146:157–162

Lafountain KL, Mascarenhas JP (1972) Isolation of vegetative

and generative nuclei from pollen tubes. Exp Cell Res

73:233–236. doi:10.1016/0014-4827(72)90125-5

Lamote V, Baert J, Roldan-Ruiz I, De Loose M, Van Bock-

staele E (2002) Tracing of 2n egg occurence in perennial

ryegrass (Lolium perenne L.) using interploidy crosses.

Euphytica 123:159–164. doi:10.1023/A:1014980123519

Lim K-B, Chung J-D, Van Kronenburg BCE, Ramanna MS,

De Jong JH, Van Tuyl JM (2000) Introgression of Liliumrubellum Baker chromosomes into L. longiflorum Thunb.:

a genome painting study of the F1 hybrid, BC1 and BC2

progenies. Chromosome Res 8:119–125. doi:10.1023/A:

1009290418889

Murashige T, Skoog F (1962) A revised medium for rapid growth

and bioassays with tobacco tissue cultures. Physiol Plant

15:473–497. doi:10.1111/j.1399-3054.1962.tb08052.x

Negri V, Lemmi G (1998) Effect of selection and temperature

stress on the production of 2n gametes in Lotus tenuis.

Plant Breed 117:345–349. doi:10.1111/j.1439-0523.1998.

tb01952.x

Oginuma K, Peng CI (2002) Karyomorphology of Taiwanese

Begonia: taxonomic implications. J Plant Res 115:225–

235. doi:10.1007/s102650200028

Pagliarini MS, Takayama SY, de Freitas PM, Carraro LR,

Adamowski EV, Silva N, Batista LAR (1999) Failure of

cytokinesis and 2n gamete formation in Brazilian acces-

sions of Paspalum. Euphytica 108:129–135. doi:

10.1023/A:1003660327223

Pan G, Zhou Y, Fowke LC, Wang H (2004) An efficient

method for flow cytometric analysis of pollen and detec-

tion of 2n nuclei in Brassica napus pollen. Plant Cell Rep

23:196–202. doi:10.1007/s00299-004-0830-y

Ramanna MS, Jacobsen E (2003) Relevance of sexual poly-

ploidization for crop improvement—a review. Euphytica

133:3–18. doi:10.1023/A:1025600824483

Ramsey J, Schemske DW (1998) Pathways, mechanisms and

rates of polyploidy formation in the flowering plants.

Annu Rev Ecol Syst 29:267–501. doi:10.1146/annurev.

ecolsys.29.1.467

Russell SD (1991) Isolation and characterisation of sperm cells

in flowering plants. Annu Rev Plant Physiol Plant Mol Biol

42:189–204. doi:10.1146/annurev.pp.42.060191.001201

Sala CA, Camadro EL, Salaberry MT, Mendiburu AO (1989)

Cytological mechanism of 2n pollen formation and unilat-

eral sexual polyploidization in Lolium. Euphytica 43:1–6.

doi:10.1007/BF00037889

Tebbitt MC (ed) (2005) Begonias: cultivation, identification

and natural history. Timberpress, Portland

Euphytica (2009) 168:81–94 93

123

Trojack-Goluch A, Berbec A (2003) Cytological investigations

of the interspecific hybrids of Nicotiana tabacum L. 9 N.glauca Grah. J Appl Genet 44:45–54

Van Breukelen EWM (1982) Competition between 2x and x

pollen in styles of Solanum tuberosum determined by a

quick in vivo method. Euphytica 31:585–590. doi:

10.1007/BF00039196

Van Tuyl JM, de Vries JN, Bino RJ, Kwakkenbos AM (1989)

Identification of 2n pollen producing interspecific hybrids

of Lilium using flow cytometry. Cytologia (Tokyo)

54:737–745

Vaughn K, Lehnen L (1991) Mitotic disruptor herbicides.

Weed Sci 39:450–457

Veilleux R (1985) Diploid and polyploid gametes in crop

plants: mechanisms of formation and utilization in plant

breeding. Plant Breed Rev 3:41–46

94 Euphytica (2009) 168:81–94

123