Sex Plant Reprod (2001) 14:195–200
DOI 10.1007/s00497-001-0118-0
O R I G I N A L A RT I C L E
Tamara N. Naumova · Jessica van der Laak
Jaroslaw Osadtchiy · Fritz Matzk · Alexei Kravtchenko
Jan Bergervoet · Kamisetti S. Ramulu · Kim Boutilier
Reproductive development in apomictic populations
of Arabis holboellii (Brassicaceae)
Received: 25 April 2001 / Accepted: 19 October 2001 / Published online: 16 November 2001
© Springer-Verlag 2001
Abstract Megasporogenesis, megagametogenesis and seed
formation were analyzed cytologically in populations of
Arabis holboellii originating from North America (Colora-
do) and Greenland. The Colorado population contained
only triploid plants, while the Greenland population con-
sisted of diploid and triploid plants. The penetrance of the
apomictic trait was assessed at the level of embryo sac de-
velopment. All populations showed facultative apomeiotic
embryo sac development; however the penetrance of this
trait differed between the populations. Apomeiotic and
meiotic embryo sac development were characterized
by diplosporous dyad formation (Taraxacum-type) and
meiotic tetrad formation (Polygonum-type), respectively.
Flow cytometric analyses of single mature seeds from all
three populations suggest that only unreduced gametes
participate in viable seed development. Pseudogamy was
the predominant mode of endosperm formation; however,
autonomous endosperm development was also observed.
The fertilization of unreduced egg cells with unreduced
pollen was observed at a low frequency in the Greenland
populations. The mechanisms of apomictic reproduction
in A. holboellii are discussed.
Keywords Arabis holboellii · Apomixis · Diplospory ·
Autonomous endosperm · Pseudogamy
T.N. Naumova, J. van der Laak and J. Osadtchiy contributed
equally to this work
T.N. Naumova · J. van der Laak · J. Osadtchiy · A. Kravtchenko
J. Bergervoet · K.S. Ramulu · K. Boutilier (✉)
Plant Research International, Postbus 16, 6700 AA, Wageningen,
The Netherlands
e-mail: k.a.boutilier@plant.wag-ur.nl
Tel.: +31-317-477167, Fax: +31-317-423110
T.N. Naumova · J. Osadtchiy
Komarov Botanical Institute, St. Petersburg, Russia
F. Matzk
Institut für Pflanzengenetik und Kulturpflanzenforschung (IPK),
Gatersleben, Germany
A. Kravtchenko
Institute for Cytology and Genetics, Novosibirsk, Russia
Introduction
Apomixis in plants is a natural phenomenon whereby
seeds are produced with embryos that are genotypically
identical to the maternal parent. Apomictic reproduction
is characterized by the avoidance or modification of
meiosis leading to unreduced gametes, parthenogenetic
embryo development and pseudogamous, or in some
cases, autonomous endosperm development (Nogler
1984; Asker and Jerling 1992).
We are using Arabis holboellii as a model to under-
stand the developmental and molecular regulation of
apomixis in plants. A. holboellii is the only reported apo-
mict among the Brassicaceae; however, a number of
Arabis species have ploidy levels (e.g. 3×) that suggest
they also reproduce by apomixis (Mulligan 1995; Koch
et al. 1999; Sharbel and Mitchell-Olds 2001). Sexual and
apomictic development in A. holboellii have been de-
scribed cytologically in great detail by Böcher (1951).
Evidence for apomixis at the molecular level has also
been described (Roy and Rieseberg 1989; Roy 1995).
A. holboellii has several characteristics that make it
an attractive model system to study apomixis. It is a
close relative of Arabidopsis thaliana; therefore the ex-
tensive molecular genetic resources that exist for A. tha-
liana – including the complete genome sequence and
wide range of reproductive mutants – can be used to
identify and isolate genes involved in apomictic process-
es in A. holboellii. Further, A. holboellii is a highly poly-
morphic species, including both sexual and apomictic
forms that vary in ploidy and geographic origin (Rollins
1941; Böcher 1947, 1954; Sharbel and Mitchell-Olds
2001), providing a source of natural variation for genetic
studies on the regulation of apomictic development.
As a first step toward developing A. holboellii as a
model system, we have used cytological techniques to
characterize megasporogenesis, embryo sac and seed
development in populations originating from North
America and Greenland. We confirm the work of Böcher
(1951) by describing A. holboellii as a diplosporous apo-
mict showing apomixis at both the diploid and triploid
196
level. Further, we document the occurrence of hybrids
resulting from unreduced female and male gametes, as
well as pseudogamous and autonomous endosperm de-
velopment in this species.
Materials and methods
Plant material
Seeds of A. holboellii were originally collected from plants grow-
ing in natural habitats of North America (Colorado) and Green-
land, and were obtained from Bitty Roy (Swiss Federal Institute of
Technology) via Rod Scott (University of Bath) and from Bettina
Lehnhardt (Humbolt University), respectively. The plants were
maintained in the greenhouse and were grown under long-day
conditions on sandy soil mixed with small clay rocks.
Ovule and seed development
For the study of ovule and seed development, whole ovaries and
siliques were fixed in either FAA or Carnoy’s solution and cleared
as in Naumova et al. (1999). The ovules and seeds from individual
ovaries and siliques, respectively, were mounted on separate slides
for further investigation. Multiple ovaries/siliques from different
branches of the same plant were analyzed for each plant. Cleared
ovules and seeds were investigated with phase contrast and
Nomarski optics. Embryo sacs were scored as meiotic or apome-
iotic based on the presence of a tetrad or a dyad with a developing
chalazal megaspore, respectively. Semi-thin sections (4 µm) were
prepared from fixed, Technovit (Heraeus Kulzer)-embedded mate-
rial and stained with toluidine blue.
Ploidy and pollen analysis
Chromosome numbers were determined in microsporocytes by
staining Carnoy-fixed anthers with DAPI solution. Ploidy levels of
leaf cells were determined using flow cytometry. Suspensions of
intact leaf nuclei were prepared (Arumuganathan and Earle 1991;
Naumova et al. 1999) and analyzed using a Beckman-Coulter
EPICS XL-MCL flow cytometer (Miami, Fla.) equipped with an
argon ion laser at 488 nm. The ploidy of embryo and endosperm
nuclei was determined in single mature seeds using a Partec Flow
Cytometer PA (Münster, Germany) and the Flow Cytometric Seed
Screen (FCSS; Matzk et al. 2000, 2001). Pollen viability was esti-
mated by Alexander staining (Alexander 1969).
Results
Arabis holboellii populations
The Colorado population was obtained as bulked seeds
from which 78 plants were grown. All plants showed a
uniform and robust phenotype (Fig. 1A). Flowering be-
gan 6 months after sowing, lasted for approximately
6 months, and could be extended to more than 1 year by
removing the main inflorescence stem (Table 1). Both
pollen viability, as measured by Alexander staining, and
seed set were generally high; however both were low at
the beginning and end of the flowering period. Chromo-
some counts and ploidy analysis of 11 plants (Table 1)
showed that these plants are triploid (2n = 3× = 21;
Fig. 1B). This population is referred to hereafter as ‘trip-
loid Colorado’.
The Greenland population was obtained as eight seed
batches among which two phenotypes were observed.
Fig. 1A–E Representative phenotypes and chromosome numbers
of the Arabis holboellii populations. A Triploid Colorado plant.
B Microsporocyte at metaphase I from a triploid Colorado plant.
Note the 21 univalent chromosomes. Triploid plants produce
mostly unreduced pollen. ×2,200. C Diploid Greenland plant.
D Triploid Greenland plant. E Microsporocyte at metaphase I
from a diploid Greenland plant showing 14 univalent chromo-
somes. Diploid plants produce a mixture of reduced and unre-
duced pollen in a single anther. ×2,200
 197
Table 1 Characteristics of the Arabis holboellii populations used in this study

Population Ploidy (number of Time to flowering Duration of Pollen fertility (%)

plants analysed) (months) flowering (months)
Mean Range

Triploid Colorado 3× (11) 6 6–12 95 50–100

Diploid Greenland 2× (15) 5 3–4 33 0–98
Triploid Greenland 3× (11) 6 6–12 50 0–95

Table 2 Reconstruction of reproductive pathways in A. holboellii ovules and mature seedsa

Population Female gametophyte development Seed development

No. plants No. ovules % Apomeiosis No. plants C values

analyzed analyzed analyzed embryo:endosperm
Average Range (number of seeds)

Triploid Colorado 4 671 98 95–100 4 3C:9C (40)b

3C:6C (5)c
Diploid Greenland 16 462 73 45–89 5 2C:6C (55)b
2C:4C (1)c
4C:6C (1)d
2.6C:6C (1)e
Triploid Greenland 4 513 94 83–98 2 3C:9C (21)b
3C:6C (1)c
6C:9C (1)d
a For female gametophyte development, ovules were cleared and b Unreduced embryo sac, egg cell parthenogenesis and pseudo-
then scored for meiotic (tetrads) or apomeiotic (dyads with a de- gamous endosperm development (fertilization of the central cell
veloping chalazal megaspore) embryo sacs. The ‘% apomeiosis’ with unreduced pollen)
was calculated per plant based on the total number of apomeiotic c Unreduced embryo sac, egg cell parthenogenesis and autono-
and meiotic embryo sacs observed for that plant. The mean was mous endosperm development
calculated as the sum of the per plant ‘% apomeiosis’ values di- d Unreduced embryo sac and double fertilization by unreduced
vided by the total number of plants observed. Reproductive path- pollen
ways in seeds were determined using Flow Cytometric Seed e Unreduced embryo sac (possibly aneuploid), parthenogenesis
Screen (FCSS) analysis on single mature seeds and pseudogamous endosperm (fertilization of the central cell
with unreduced pollen)

One group of plants (five seed batches, 39 plants grown)
developed as a slender rosette from which multiple inflo-
rescence stems emerged (Fig. 1C). These plants flowered
5 months after sowing, and continued to flower for an
additional 3–4 months (Table 1). Cutting back the inflo-
rescence stems did not extend the flowering period. Ap-
proximately one third of these plants were completely
male and female sterile. The remaining plants showed
variable male and female sterility on separate branches
of the same plant. Chromosome counts of 15 of the 39
plants (Table 1) showed that these plants were diploid
(2n = 2× = 14; Fig. 1E). This group of plants is hereafter
referred to as ‘diploid Greenland’.
The second phenotypic group of Greenland plants
(three seed batches, 24 plants grown) developed as a sin-
gle robust rosette carrying one to three inflorescence
stems (Fig. 1D). Chromosome counts and ploidy analy-
sis of 11 of the 24 plants showed that these plants were
triploid (2n=3×= 21). This group of plants is referred to
as ‘triploid Greenland’. The timing and period of flower-
ing, as well as the pollen fertility and seed set character-
istics are as described above for the triploid Colorado
population (Table 1), with the exception that three of the
24 plants were completely sterile.
Megasporogenesis and embryo sac development
Megasporogenesis and megagametogenesis were investi-
gated in all three populations using tissue clearing. The
pathway of embryo sac development (apomeiotic or
meiotic) was investigated to determine the mode of re-
production and to estimate the percentage of apomeiotic
ovules per plant. The results of these analyses are pre-
sented below and are summarized in Table 2.
Embryo sac development began with the differentia-
tion of the archesporial cell, at the stage when the inner
and outer integuments began to develop. The archespori-
al cell followed two different modes of development. In
70% of all ovules examined, the archesporial cell divid-
ed to produce a megaspore mother cell (MMC) and a pa-
rietal cell. In the remaining ovules, the archesporial cell
developed directly into the MMC (Fig. 2A). The subse-
quent development of the MMC differed between the
meiotic and apomeiotic pathways. In the meiotic path-
way, the MMC divided twice to form a tetrad of mega-
spores (Fig. 2F). The three micropylar megaspores of the
tetrad degenerated, while the chalazal megaspore en-
larged and then underwent three mitoses to form an
eight-nucleate Polygonum-type embryo sac. In the apo-
198
Fig. 2A–L Reproductive pathways in triploid and diploid A. hol-
boellii populations. Images A, E, G, J are from triploid Colorado
plants, H, I from triploid Greenland plants and B–D, F, K, L from
diploid Greenland plants. Images A–D and F–K are from cleared
ovules. Images E and L are from sectioned material. A Ovule with
megaspore mother cell (MMC) in a subepidermal position. The in-
ner (II) and outer (OI) integuments have almost enclosed the nu-
cellus (N). ×660. B Ovule with a diplosporous dyad (Dy). ×760.
C As in B, but with a parietal cell (PC) above the dyad. ×900.
D Ovule with a one-nucleate diplosporous Taraxacum-type em-
bryo sac (DES) and remnants of the micropylar megaspore-like
cell (‘‘MS’’). ×790. E As in D, but with a subepidermal parietal
cell (PC). ×1,000. F Tetrad of meiotic megaspores (TMS). The en-
larged chalazal megaspore is functional (FMS). A parietal cell
(PC) is present. ×670. G Ovule with a fully developed Taraxacum-
type embryo sac showing an egg cell (EC), two synergids (SY) and
a central cell with two polar nuclei (PN). ×370. H Ovule with em-
bryo sac showing the two intact synergids (SY), the egg cell (EC)
and the primary endosperm nucleus (En) formed after fusion of
the polar nuclei. ×360. I Initial steps of free nuclear endosperm
(En, open arrowheads) development. The egg cell (EC) is elongat-
ed. ×330. J Seed with a globular-stage embryo and multinuclear
endosperm (En). ×150. K Seed with globular stage embryo (E)
and a restricted amount of endosperm (En). A large hypostase (H)
is visible. ×100. L Cross section of a seed containing an early cot-
yledon-stage embryo and endosperm. The micropylar endosperm
has cellularized, while the chalazal endosperm is still nuclear. ×57
meiotic pathway, the MMC divided once to produce a
diplosporous dyad consisting of two unreduced mega-
spore-like cells (“MS”; Fig. 2B, C). The micropylar
“MS” of the diplosporous dyad degenerated, while the
chalazal “MS” increased in size, underwent vacuoliza-
tion (Fig. 2D, E) and three subsequent mitoses to form
an eight-nucleate embryo sac (Taraxacum-type). The ma-
ture Taraxacum-type embryo sac, which is morphologi-
cally indistinguishable from a meiotic embryo sac
(Fig. 2G). Both meiotic and apomeiotic embryo sac de-
velopment were observed in ovules from the same ovary
of diploid Greenland plants. The frequency of apomeio-
tic embryo sac formation in diploid Greenland plants
was estimated to range from 45 to 89% (Table 2).
Seed development
Apomictic seed development was morphologically simi-
lar in all three populations. Soon after anthesis, the apo-
meiotic embryo sac increased greatly in size and the polar
nuclei fused and moved to a central position within the
central cell (Fig. 2H). Nuclear endosperm development
was initiated prior to embryo formation (Fig. 2I). Subse-
quent embryo and endosperm development (Fig. 2J, L),

were as described for sexual members of the Brassicaceae
(Maheshwari 1950; Brown et al. 1999), except that seeds
with a restricted endosperm content were observed in
both the Greenland (Fig. 2K) and Colorado populations.
Seeds with a restricted amount of endosperm were ap-
proximately the same size as seeds in the same silique
with a normal endosperm content.
We used FCSS (Matzk et al. 2000, 2001) to more pre-
cisely determine the pathway of embryo and endosperm
development in mature A. holboellii seeds. All three pop-
ulations were comparable in their reproductive behavior
(Table 2). Pseudogamy (fertilization of the central cell)
was the predominant mode of endosperm formation in
all three populations; however, autonomous endosperm
development was occasionally observed. This data is in
agreement with the results of Matzk et al. (2000) who
demonstrated both autonomous and pseudogamous endo-
sperm development using bulked seeds for the flow cyto-
metric analysis of obligate and facultative A. holboellii
apomicts. Fertilization of unreduced egg cells and pseu-
dogamous endosperm were also detected in the Green-
land populations. Interestingly, preliminary flow cyto-
metric analyses of single mature seeds from the diploid
Greenland population suggest that only unreduced ga-
metes contribute to the formation of mature seeds (ab-
sence of 2:3 or 3:5 embryo:endosperm C-values).
Discussion
Reproductive development in A. holboellii was initially
described by Böcher (1951). His detailed observations,
based mainly on laborious serial sectioning, attracted our
attention and served as the starting point for our present
investigation. Our results are in general agreement with,
and extend, Böcher’s early observations (Böcher 1951).
The diploid A. holboellii population is exceptional in
that the majority of reported gametophytic apomicts are
polyploids (Asker and Jerling 1992). A number of hy-
potheses have been put forward to account for the rare
occurrence of apomixis at the diploid level (Nogler
1984; Mogie 1992; Carman 1997). Carman’s hybridiza-
tion-derived floral asynchrony theory (Carman 1997)
states that apomixis does not arise from expression of
specific apomixis alleles, but rather from asynchronous
expression of duplicate sexual reproductive gene sets in
hybrid/polyploid genomes. These hybrids are often
Pleistocene relicts with disjunctive ranges that have sur-
vived numerous glacial and interglacial periods, as may
be the case for A. holboellii. Diploid apomictic A. hol-
boellii populations could have retained functional copies
of duplicate asynchronously expressed reproductive gene
sets during their descent from polyploidy to diploidy.
Nogler (1984) suggests that natural diploid apomicts
are rare because the alleles responsible for apomixis may
act as, or are linked to, recessive lethal factors and can
therefore only be transferred via a diploid or polyploid
gamete. The A. holboellii diploids produce both meiotic
and apomeiotic embryo sacs within the same ovary.
199
Male meiosis is likewise affected in that a low percent-
age of pollen mother cells undergo apomeiotic develop-
ment (see Fig. 1E), leading to a mixture of reduced
and unreduced pollen grains in a single anther (A.
Kravtchenko, unpublished results). Our preliminary
FCSS data suggest that in diploid plants only unreduced
female and male gametes contribute to mature seed for-
mation (in all cases of parthenogenetic and zygotic em-
bryo development, as well as pseudogamous endosperm
formation). However, analysis of megasporogenesis and
megagametogenesis in the diploid population indicated a
significant percentage of meiotic embryo sac develop-
ment. Taken together, these results suggest that there is
selection against, or developmental abnormalities associ-
ated with, haploid gametes. Additional studies on a larg-
er number of plants and a wider range of populations are
clearly required.
The work described here serves as a basis for future
studies aimed at the identification and characterization of
the genes controlling apomictic development in A. hol-
boellii. A recent analysis demonstrated that Capsella ru-
bella, a species in the same clade as A. holboellii, shares
extensive genome colinearity and sequence similarity
with A. thaliana (Acarkan et al. 2000). Gene sequencing
efforts in our group indicate that many A. holboelli
cDNA sequences show greater than 90% DNA sequence
similarity with those from A. thaliana. This important
characteristic allows us to make use of the wide variety
of molecular tools developed for A. thaliana to acceler-
ate the mapping and identification of apomictic genes in
A. holboellii.
Acknowledgements We thank P. Dijkhuis for technical assis-
tance, R. Scott and B. Lehnhardt for A. holboellii seeds, J. van Oss
for photographic work, T. Sharbel for advice on performing chro-
mosome counts, C.-M. Liu for helpful discussions and two anony-
mous reviewers for constructive comments on the manuscript. The
research was supported by NWO Russian-Dutch Research Coop-
eration grant number 047.007.019.
References
Acarkan A, Rossberg M, Koch M, Schmidt R (2000) Comparative
genome analysis reveals extensive conservation of genome or-
ganisation for Arabidopsis thaliana and Capsella rubella.
Plant J 23:55–62
Alexander MP (1969) Differential staining of aborted and non-
aborted pollen. Stain Technol 44:117–122
Arumuganathan K, Earle ED (1991) Estimation of nuclear DNA
content of plants by flow cytometry. Plant Mol Biol Rep
9:229–233
Asker SE, Jerling L (1992) Apomixis in plants. CRC Press, Boca
Raton, Fla.
Böcher TW (1947) Cytological studies of Arabis holboellii. Here-
ditas 32:573–575
Böcher TW (1951) Cytological and embryological studies in the
amphi-apomictic Arabis holboellii complex. Kong Dansk
Vidensk Selsk Biol Skr VI 7:1–57
Böcher TW (1954) Experimental taxonomical studies in the Ara-
bis holboellii complex. Sven Bot Tidskr 48 H 1:31–44
Brown RC, Lemmon BE, Nguyen H, Olsen O-A (1999) Develop-
ment of endosperm in Arabidopsis thaliana. Sex Plant Reprod
12:32–42
200
Carman JG (1997) Asynchronous expression of duplicate genes in
angiosperms may cause apomixis, bispory, tetraspory and
polyembryony. Biol J Linn Soc 61:51–94
Koch M, Bishop J, Mitchell-Olds T (1999) Molecular systematics
and evolution of Arabidopsis and Arabis. Plant Biol 1:529–
537
Maheshwari P (1950) An introduction to the embryology of angio-
sperms. McGraw-Hill, New York.
Matzk F, Meister A, Schubert I (2000) An efficient screen for re-
productive pathways using mature seeds of monocots and di-
cots. Plant J 21:97–108
Matzk F, Meister A, Brutovska R, Schubert I (2001) Reconstruc-
tion of reproductive diversity in Hypericum perforatum L.
opens novel strategies to manage apomixis. Plant J 26:275–
82
Mogie M (1992) The evolution of asexual reproduction in plants.
Chapman and Hall, London
Mulligan GA (1995) Synopsis of the genus Arabis (Brassicaceae)
in Canada, Alaska and Greenland. Rhodora 97:109–163
Naumova TN, Hayward D, Wagenvoort M (1999) Apomixis and
sexuality in diploid and tetraploid accessions of Brachiaria
decumbens. Sex Plant Reprod 12:43–52
Nogler GA (1984) Gametophytic apomixis. In: Johri BM (ed) Em-
bryology of angiosperms. Springer, Berlin Heidelberg New
York, pp 475–518
Rollins RC (1941) A monographic study of Arabis in Western
North America. Rhodora 43:289–315
Roy BA (1995) The breeding systems of six species of Arabis
(Brassicaceae). Am J Bot 82:869–877
Roy BA, Rieseberg LH (1989) Evidence for apomixis in Arabis.
J Hered 80:506–507
Sharbel TF, Mitchell-Olds T (2001) Recurrent polyploid origins
and chloroplast phyleogeography in the Arabis holboellii com-
plex (Brassicaceae). Heredity 86:59–68