Between sexual and apomictic: unexpectedly variable sporogenesis and production of viable polyhaploids in the pentaploid fern of the Dryopteris affinis agg. (Dryopteridaceae)

Abstract

Background and Aims In ferns, apomixis is an important mode of asexual reproduction. Although the mechanisms of fern reproduction have been studied thoroughly, most previous work has focused on cases in which ferns reproduce either exclusively sexually or exclusively asexually. Reproduction of ferns with potentially mixed systems and inheritance of apomixis remains largely unknown. This study addresses reproduction of the pentaploid Dryopteris × critica, a hybrid of triploid apomictic D. borreri and tetraploid sexual D. filix-mas.

Methods Spore size, abortion percentage and number of spores per sporangium were examined in pentaploid plants of D. × critica grown in an experimental garden. The sporangial content of leaf segments was cultivated on an agar medium, and DNA ploidy levels were estimated by DAPI flow cytometry in 259 gametophytes or sporophytes arising from the F₂ generation of the pentaploid hybrid.

Key Results The hybrid is partly fertile (89–94 % of aborted spores) and shows unstable sporogenesis with sexual and apomictic reproduction combined. The number of spores per sporangium varied from approx. 31 to 64. Within a single sporangium it was possible to detect formation of either only aborted spores or various mixtures of aborted and well-developed reduced spores and unreduced diplospores. The spores germinated in viable gametophytes with two ploidy levels: pentaploid (5x, from unreduced spores) and half of that (approx. 2·5x, from reduced spores). Moreover, 2–15 % of gametophytes (both 2·5x and 5x) formed a viable sporophyte of the same ploidy level due to apogamy.

Conclusions This study documents the mixed reproductive mode of a hybrid between apomictic and sexual ferns. Both sexual reduced and apomictic unreduced spores can be produced by a single individual, and even within a single sporangium. Both types of spores give rise to viable F₂ generation gametophytes and sporophytes.

INTRODUCTION

While sexual reproduction is a process that creates a new genetic entity by combining the genetic material of two parental individuals, asexual reproduction is confined to one genetic entity and maintains its integrity even in the case of imperfect reproductive isolation from other entities. Apomixis produces progeny asexually by different means both in flowering plants (Ozias-Akins, 2006; Krahulcová et al., 2013) and in ferns (Döpp, 1939; Manton, 1950). Among ferns, apomixis evolved several times independently and its frequency is at least 3 %, a value much higher than in other major plant groups (Liu et al., 2012). However, most apomictic fern species are concentrated in just four families (Liu et al., 2012).

Apomixis in ferns includes apogamy – the formation of sporophytes from somatic cells of the prothallium – and agamospory – the production of unreduced (diplo)spores (Manton, 1950; Lovis, 1977; Walker, 1979; Gastony and Windham, 1989). The archesporial cell of sexual fern species usually undergoes four mitoses to produce 16 spore mother cells that undergo regular meiosis, resulting in 64 reduced spores in 16 tetrads. Under the prevailing type of agamospory (Döpp–Manton scheme) the last (premeiotic) mitosis fails, resulting in eight spore mother cells that undergo regular meiosis, producing 32 diplospores in eight tetrads (Döpp, 1939; Manton, 1950; Walker, 1979). Rarely (Braithwaite scheme), the first meiotic division fails, which results in 32 diplospores in 16 diads (Braithwaite, 1964). Unlike in angiosperms, regular meiosis is present under the Döpp–Manton type of agamospory. Homologous pairing and crossing-over are thus present and were recently recognized as the possible mechanisms of formation of genetically different spores (Lin et al., 1992). Genetic variation among apomictic offspring has been documented (Ishikawa et al., 2003; Schneller and Krattinger, 2010; Ootsuki et al., 2012). In contrast to flowering plants, the fern apomicts are obligate (Lovis, 1977). The only reported case of facultative apomixis among ferns, Asplenium hallbergii, remains under study (Dyer et al., 2012). Autopolyploidy, hybridization or fusion of reduced and unreduced gametes may play a role in formation in the apomictic polyploid ferns (Barrington et al., 1989; Park and Kato, 2003; Grusz et al., 2009; Hunt et al., 2011; Chao et al., 2012; Liu et al., 2012).

In sexual fern hybrids, an abnormal meiosis yields variable percentages of aborted (non-viable) or atypical spores of different size and shape from regular spores (Wagner and Chen, 1965; Gabriel y Galán and Prada, 2011; Zhang et al., 2013) and these hybrids are sterile or nearly so (Wagner and Chen, 1965; Reichstein, 1981).

Hybridization of fern apomicts with related sexual taxa can give rise to new fertile apomictic taxa of higher ploidy levels via diplospores of the apomictic parent (Gastony and Windham, 1989; Fraser-Jenkins, 2007; Grusz et al., 2009; Regalado Gabancho et al., 2010; Dyer et al., 2012). The prothallia of apomictic ferns normally lack functional archegonia but may possess functional antheridia, releasing unreduced spermatozoids that are capable of fertilizing the archegonia of sexual species. It is believed that the resulting hybrids inherit the apomictic mode of reproduction from their male parents (Döpp, 1955; Walker, 1979; Gastony and Windham, 1989; Windham and Yatskievych, 2003; Regalado Gabancho et al., 2010; Liu et al., 2012). However, some studies have reported mixed meiosis in these hybrids with joint existence of eight-celled (apomictic) and 16-celled (sexual-like) meiosis in one plant (Schneller, 1975; Dyer et al., 2012). Thus, spore formation and offspring constitution and viability remain unresolved in the sexual × apomictic fern hybrids.

In Europe, the most thoroughly studied group was the apomictic complex Dryopteris affinis agg. (Manton, 1950; Döpp, 1955; Schneller, 1975; Fraser-Jenkins, 2007; Bär and Eschelmüller, 2010; Schneller and Krattinger, 2010). In Central Europe it consists of diploid D. affinis and triploid D. borreri and D. cambrensis (see Ekrt et al., 2009, for ploidy levels and genome sizes). The widespread sexual D. filix-mas is capable of hybridization with apomictic taxa, resulting in tetraploid and pentaploid hybrids. Hybrids form both aborted and well-developed spores (Schneller, 1975; Eschelmüller, 1998; Fraser-Jenkins, 2007; Ekrt et al., 2009; Bär and Eschelmüller, 2010) that are able to germinate. Manton (1950), in her famous study of fern cytology and reproduction, described experiments with ‘pentaploid D. borreri’ (i.e. D. × critica) in which she observed apparently functional spores beside aborted ones and germination of gametophytes from these spores, including probably also those from sexual-like sporangia with 16 spore mother cells. However, no further information on these offspring is available (and could hardly be so using the methods of that time). Schneller (1975) also reported the occurrence of karyologically variable and mostly aneuploid offspring (gametophytes) of the pentaploid D. × critica.

Our case study follows the above studies. We focused on fertility and offspring viability of the pentaploid hybrid Dryopteris × critica (2n = 205), which is the hybrid between triploid apomictic D. borreri (2n = 123) and tetraploid sexual D. filix-mas (2n = 164). Here, we attempt to answer the following questions: (1) What is the portion of aborted and viable spores? (2) Does the pentaploid F₁ hybrid produce viable F₂ offspring, and if so, are any offspring gametophytes able to form sporophytes? (3) What is the pattern of genome size/ploidy levels among maternal plants and gametophytes/sporophytes arisen from spores of the F₁ pentaploid hybrid? (4) What is the frequency of hybridization at sites of common occurrence of parental taxa D. borreri and D. filix-mas in the wild?

METHODS

Spore size and abortion

During a previous study of Dryopteris affinis agg. in Central Europe (Ekrt et al., 2009), only a few pentaploid plants (Dryopteris × critica) were detected in the wild. Two plants from different locations (STO, KUR, see Table 1) were transplanted into the experimental garden. To avoid contamination with fern spores from the surrounding area, one leaf of each plant was enveloped with several layers of UHELON 130T Extra textile, 25-µm mesh size (Silk & Progress, s.r.o., Brněnec, Czech Republic). The mature fertile fronds were collected at the start of spontaneous snapping, wrapped in paper sheets and dried at room temperature to release sporangial contents. Spore size and abortion percentage of the experimental plants were studied to estimate spore fitness. Spores were investigated under a light microscope (Olympus CH30) at 400× magnification. The spore abortion percentage was estimated in a random sample of 1000 spores per plant. Spores were considered to be aborted when they lacked the protoplast or were collapsed (Quintanilla and Escudero, 2006). Exospore length was examined in a random sample of 200 well-shaped spores at 1000× magnification. The central part of a frond bearing ripe and still undehisced sporangia was fixed in 50 % ethanol. Under the light microscope, a sporangium was opened by a thin needle in a drop of water and sporangial content was examined. The number of spores (including aborted spores) per sporangium and exospore length of well-developed spores were recorded in 15 separate sporangia.

Table 1.

Localization of plants used in the study; herbarium vouchers are deposited in the herbarium CBFS

Locality code	Location	Altitude (m)	Coordinates (WGS 84)	No. of plants examined
STO	Czech Republic, Šumava Mts, Stožec: beech forest in the Stožec Mt approx. 750 m E of the summit of Mt Stožec	915	48°52′55.6″N, 13°49′52.8″E	31
KUR	Slovakia, Malá Fatra Mts, Krasňany: bottom part of Kúr valley approx. 3.5 km SE of the church in the village of Krasňany	605	49°11′41.6″N, 18°55′45.5″E	143
KNE	Czech Republic, Moravskoslezské Beskydy Mts, Čeladná: massif of Kněhyně Mt, valley of Korábský stream in foothills of Malá Stolová Mt approx. 4.4 km SSW of the church in the village of Čeladná	640	49°30′39.1″N, 18°19′44.6″E	90
PEC	Czech Republic, Šumava Mts, Nová Pec: deforested line in the N slope of Smrčina Mt approx. 1.6 NNE of the summit	930	48°45′34.2″N, 13°55′50.4″E	81

Because we observed a clearly bimodal distribution of exospore length, we analysed it as a mixture of two types of spores using R 3.1.2 software (R Development Core Team, 2014). As dependence of variance on a mean and on positively skewed lognormal distributions can be expected and was apparent from preliminary analysis, we log-transformed the data. We then modelled the log-transformed data as a mixture of two Gaussian distributions using the normalmixEM function from the mixtools package (Benaglia et al., 2009).

Spore germination

Sporangial content of two leaf segments per experimental plant was poured out and cultivated in four replicates on Petri dishes (6 cm in diameter) with mineral agar BG11 (Stanier et al., 1971) at 19 °C, light intensity of approx. 50 µE and 16/8 h light–dark. The dishes were sealed with Parafilm to reduce contamination and prevent excessive water loss. Young gametophytes were transplanted into Petri dishes with a sterilized peat/sand mixture (3 : 1) and were placed approx. 0·5 cm from one another. During 5 months of cultivation, well-developed gametophytes and young sporophytes (if present) were examined by flow cytometry (FCM).

Screening of wild populations

Localities of both experimental plants (STO, KUR) and two other localities (KNE, PEC) were screened for genome size variation using FCM (Table 1). At each locality, the study plot of approx. 100 × 150 m was established and leaves of all plants with D. affinis agg. morphology and juvenile (undeterminable) individuals were collected. We also collected a smaller number of individuals of D. filix-mas, which is dominant in the localities and their surroundings. The leaves were stored moist in plastic bags up to 4 d for FCM analyses. Voucher specimens are stored in the herbarium CBFS.

Flow cytometry

Relative DNA content and DNA ploidy levels were determined using a Partec PA II flow cytometer (Partec GmbH., Münster, Germany) equipped with a mercury arc lamp. Fresh material was analysed (field-collected leaves or cultivated gametophytes/sporophytes). Samples were prepared following the simplified two-step protocol of Doležel et al. (2007). For adult leaves, approx. 2 cm² of intact leaf tissue was chopped with a sharp razor blade together with approx. 0·25 cm² of an internal standard leaf (Vicia faba ‘Inovec’, 2C = 26·90 pg; Doležel et al., 1992) in a plastic Petri dish containing 0·5 mL of ice-cold Otto I buffer (0·1 m citric acid, 0·5 % Tween-20). The suspension was filtered through a 42-µm nylon mesh and incubated for at least 5 min at room temperature. After incubation, 1 mL of the staining solution was added. The staining solution consisted of 1 mL of Otto II buffer (0·4 m Na₂HPO₄.12H₂O), 2-mercaptoethanol (2 µL mL^–1) and the fluorochrome DAPI (4 µL mL^–1). Samples were run on the flow cytometer after approx. 1 min of staining and the fluorescence intensity of 3000–5000 particles was recorded. For screening of ploidy levels, pooled samples of up to five individuals could be used as we utilized high-resolution histograms and owing to the absence of endopolyploidy. Nevertheless, each plant was separately re-analysed if the occurrence of more DNA ploidy levels in the pooled sample was suspected. For gametophytes and young sporophytes, the same method was used but the amount of available plant material was much smaller. Only large well-developed gametophytes were measurable (yielded enough nuclei). We used the whole gametophyte/one young leaf of a sporophyte and small amount (<2 × 2 mm) of the internal standard. Although we ran the whole volume of the sample, we were usually not able to record the usual 3000 particles per sample even with the largest gametophytes; however, the scored peaks were clear and included at least several hundred nuclei. We did not use pooled samples for gametophytes/young sporophytes.

For calibration, cultivated individuals of triploid Dryopteris borreri and tetraploid D. filix-mas with known chromosome counts (Ekrt et al., 2009) were analysed. The fluorescence histograms were evaluated using FloMax 2.6 software provided by Partec.

Differences between mean DNA contents of two groups within one DNA ploidy level were compared using t-tests with separate variance estimates and Welch approximation of the degrees of freedom.

RESULTS

Spore size and abortion

The majority of spores were aborted on both experimental plants of D. × critica (hereafter E_STO and E_KUR): 93·6 and 88·6 %, respectively, based on 1000 spores each (Fig. 1). The exospore length of well-developed spores showed a clear bimodal distribution (Fig. 2) in both experimental plants. Despite the relatively low number of observations (N = 200 for both plants) the statistical model shows similar values for both experimental plants: the mode of exospore length is estimated to be approx. 33 µm for the smaller spores and 47–51 µm for the larger spores (Fig. 2).

Fig. 1.

Three types of spores detected in the experimental plant E_KUR. L = large well-developed spores; S = small well-developed spores; A = aborted spores (lacking the protoplast, collapsed or of irregular shape).

Fig. 2.

Histograms of the exospore length of well-developed spores. Two size classes are apparent in each plant, and are visualized as two lognormal distributions (λ, proportion of spores that belong to the particular distribution). The parameters of the lognormal distributions were obtained from modelling a mixture of two Gaussian distributions based on log-transformed data.

Number of spores per sporangium

The content of single sporangia of the experimental plant E_STO, which produces viable reduced and unreduced spores (see below), was examined in detail. We studied 15 randomly selected sporangia. The number of spores per sporangium varied markedly from 31 to 64. The counts did not fit the textbook apomictic/sexual number of 32/64 spores per sporangium. The vast majority of spores were aborted (72·6 %). The number of well-developed spores per sporangium was variable, ranging from zero to 29. The two size classes of well-developed spores were apparent. Several different types of sporangia were detected: sporangia with all spores aborted, sporangia with a mixture of aborted and small spores or a mixture of aborted and large spores, and sporangia with a mixture of aborted, small and large spores together (Table 2, Fig. 3).

Table 2.

Summary of spore counts in single sporangia (sg) in the pentaploid plant Dryopteris × critica (E_STO); well-developed spores were classified into two size classes (see Fig. 2): small spores (exospore length 20–42 µm) and large spores (42–65 µm)

Sporangium no.	Spore count
	Total	Aborted	Small	Large
sg 1	55	55	–	–
sg 2	64	64	–	–
sg 3	60	60	–	–
sg 4	42	35	7	–
sg 5	63	40	23	–
sg 6	43	27	8	8
sg 7	47	10	8	29
sg 8	31	15	8	8
sg 9	36	23	5	8
sg 10	39	15	12	12
sg 11	51	44	–	7
sg 12	51	30	–	21
sg 13	35	26	–	9
sg 14	61	41	–	20
sg 15	53	44	–	9

Fig. 3.

Single sporangia content: (A) approx. 55 aborted spores (the sporangium sg 1); (B) a mixture of 15 aborted, eight small and eight large spores (sg 8); (C) a mixture of 26 aborted and nine large spores (sg 13); and (D) detail from B (sg 8) showing five small, one large and three aborted spores present in one sporangium.

Spore germination and sporophyte formation

All gametophytes germinated from spores of pentaploid D. × critica were filamentous at the beginning and followed the normal trend of development to the predominantly cordate phase. The vast majority of the gametophytes remained in the gametophyte stage. Only a small percentage formed antheridia and yielded sporophytes: 14·7 % (89 out of 606) in E_STO and 1·7 % (15 out of 877) in E_KUR. Gametophytes that did not form sporophytes survived for approx. 8–32 months and then died. Sporophytes were formed from the central or lower region of a gametophyte either as single ‘normal’ viable plants or rarely through a callus-like sporophytic growth (several plants originating from E_KUR). In some cases, sporophytes were deformed, having enormous pinna segmentation or split terminal leaf segments. An origin from apogamy appeared to be obligate in all the sporophytes studied.

Genome size of F₂ offspring

FCM screening of the offspring (gametophytes) of pentaploid Dryopteris × critica surprisingly revealed two cytotypes (Fig. 4, Table 3). One of them corresponded to the maternal plants and other pentaploids found in natural populations (Table 3). The other has a genome size approximately half of the pentaploids and is tentatively marked as 2·5x in this paper. The two experimental plants differed strongly in the frequency of cytotypes: in E_KUR only 2·5x offspring were found (N = 110), while in E_STO 55 % of offspring were 2·5x and 45 % were pentaploids (N = 149).

Fig. 4.

Flow cytometric profiles (DAPI staining) of F₂ gametophytes of pentaploid Dryopteris × critica: (A) 2.5x gametophyte and apogamous sporophyte analysed with the internal standard Vicia faba; (B) 5x gametophyte and apogamous sporophyte analysed with the internal standard V. faba; (C) simultaneous analysis of 2.5x F₂ gametophyte of D. × critica, triploid D. borreri and tetraploid D. filix mas (both with known chromosome count; Ekrt et al., 2009); (D) simultaneous analysis of three 2.5x gametophytes and one 5x gametophyte arisen from one maternal plant (E_STO) – three separate peaks of 2.5x gametophytes are clearly visible, corroborating variation in the genome size between the gametophytes.

Table 3.

Relative DNA content of F₂ gametophytes (type = G) and field-collected sporophytes (type = S) assessed using FCM with DAPI staining; the value is expressed as the ratio to the internal standard Vicia faba ‘Inovec’, which is given a unit value

Group	Type	N	Relative DNA content			CV (%)	Within-group variation (%)
			Mean	SE	Range
2·5x (E_KUR)	G	98	0·702	0·003	0·639–0·757	1·54–3·97	18·5
2·5x (E_STO)	G	66	0·722	0·003	0·655–0·808	1·76–3·94	23·4
3x (field)	S	266	0·885	0·001	0·869–0·900	1·08–1·90	3·6
4x (field)	S	24	1·127	0·001	1·121–1·139	1·24–1·62	1·6
5x (E_STO)	G	37	1·465	0·003	1·422–1·506	1·52–3·94	5·9
5x (field)	S	55	1·442	0·001	1·412–1·460	1·09–1·90	3·4

N, number of observations – note that some lower-quality analyses (low number of nuclei or high peak CVs) could be classified to the ploidy level but were excluded from summary statistics of the relative genome sizes; s.e., standard error of the mean; CV, coefficient of variation of the sample peak. Within-group variation describes genome size differences among samples from the respective group; it is expressed as the difference between the group maximum and minimum, which is set to 100 %.

There was considerable variation in the relative DNA content (genome size) among F₂ gametophytes (Table 3). In the 2·5x cytotype, genome size variation between gametophytes originating from one maternal plant reached 18·5 and 23·4 % in E_KUR and E_STO, respectively. The differences between individual gametophytes were corroborated also by simultaneous FCM analysis (Fig. 4D). The mean values of these two groups were also significantly different (t = 4·28, d.f. = 144·90, P = 3×10^–5). Among pentaploid gametophytes (E_STO maternal plant), variation reached 5·9 %. Although this variation is higher than among field-collected pentaploid sporophytes (3·6 %), we did not observe any bifurcated peaks in simultaneous analyses of additional gametophytes and, taking the relatively low number of nuclei (lower precision of the analyses) into account, this variation might be attributed to random measurement error. Interestingly, there was a small but significant difference between mean relative DNA contents of the experimental pentaploid gametophytes and field-collected pentaploid sporophytes (t = 4·54, d.f. = 35·08, P = 6×10^–5).

Relative genome sizes of gametophytes and sporophytes emerging from them were compared to confirm apomictic (apogamous) formation of the F₂ sporophytes. We analysed 24 gametophyte–sporophyte pairs from the E_STO experimental plant (12 of 5x ploidy level and 12 of 2·5x ploidy level) and five pairs from the E_KUR experimental plant (all of 2·5x ploidy level). In all cases the relative genome sizes were identical within the gametophyte–sporophyte pair.

Screening of the wild populations

Three ploidy levels were revealed in all four sites. Plants of D. affinis sensu lato (s.l.) morphology comprised triploids (D. borreri) and pentaploids (D. × critica). The hybrids were not always recognizable from D. borreri based on frond morphology. The proportion of hybrids among D. affinis s.l. plants was similar within three sites (KUR, KNE, PEC; 10·5–16·3 % of hybrids), but there were many more hybrids in the fourth site STO (71·4 % of hybrids). Plants of D. filix-mas morphology were all tetraploid; no tetraploid of D. affinis s.l. morphology was found. In the whole sample set of 345 plants, no 2·5x individuals were detected. Genome size variation within taxa did not exceed 3·5 % and is well within the usual random measurement error (Table 3).

DISCUSSION

Spore viability in fern hybrids

We experimentally confirmed the fertility of the pentaploid hybrid D. × critica. Both aborted and well-developed spores were detected. The spore abortion rate was approx. 89–94 %. Similar proportions of 80–95 % of aborted spores were recorded also in previous studies of the D. affinis group (Eschelmüller, 1998; Fraser-Jenkins, 2007). Eschelmüller (1998) also studied spore viability: in the pentaploid D. × critica, 80·5 % (mean from nine plants) of spores were non-viable. Similar rates of approx. 66–80 % of non-viable spores (depending on time of spore evaluation) were observed in pentaploid D. × critica by Bär and Eschelmüller (2010). A much wider scale of spore abortion rate in hybrids of sexual and apomictic taxa was detected in the genus Pteris, where the proportion of aborted spores produced by the synthetic apomictic hybrids varied from 45 to 89 % (Walker, 1962). Compared with apomictic species, hybrids between sexual fern species are either completely sterile (Reichstein, 1981; Ekrt et al., 2010) or produce only a minor proportion of viable spores (Vida and Reichstein, 1975; Pinter, 1995; Yatabe et al., 2011).

In the present study, we did not focus on spore germination/viability but rather on sporophyte formation. We also had the opportunity to estimate ploidy level using FCM, which was not available to earlier researchers. The germinated gametophytes remained mostly in the gametophyte stage but 1·7 % E_KUR (all 2·5x) and 14·7 % E_STO (both 2·5x and 5x) developed sporophytes through apogamy. The existence of viable sporophytes of the F₂ generation arisen from the pentaploid F₁ hybrid was observed for the first time. The low rate of sporophyte production may be caused both by an unbalanced number of chromosomes in gametophytes originating from reduced (approx. 2·5x) spores or suboptimal environmental conditions for germination and growth.

Spore and genome size variation

We found two size classes of spores and two ploidy levels among gametophytes. The logical explanation would be that smaller spores are reduced while larger spores are unreduced. However, this simple theory is somewhat hampered by the fact that in the experimental plant E_KUR, only reduced 2·5x gametophytes were detected, although a bimodal distribution of spore sizes was present. This result would mean that the larger (unreduced) spores were unviable for some unknown reason. The correlation between genome size and spore size has recently been challenged in the Asplenium monanthes complex (Dyer et al., 2013). This analysis was based on between-species comparisons. When phylogenetic contrasts are applied, the relationship is likely to be valid within a species or between closely related species, and was also evident from the raw data in the Asplenium monanthes complex (Dyer et al., 2013).

In our data, the genome size variation among 2·5x gametophytes was enormous. However, such a result might be expected because (1) the maternal plant is of odd-ploidy level and regular chromosome pairing in meiosis is not possible and (2) the fourth mitosis-forming restitution nuclei in Döpp–Manton type agamospory may be irregular (see below). As a result of both the problems mentioned above, many spores are aborted and even those that are well developed vary somewhat in chromosome number/genome size. Nevertheless, a small proportion of nearly balanced spores are able to germinate and some of the resultant gametophytes are even able to produce viable sporophytes. Chromosome number variation among the progeny of D. × critica was also observed by Schneller (1975).

We observed slightly higher but statistically significantly different genome size of the pentaploid gametophytes compared with more or less invariable pentaploid plants from natural populations (including the experimental maternal plant E_STO; the gametophytes are different even from this plant in a one-sample t-test). We are not aware of any mechanism that could explain such a difference; indeed, we attribute this result to technical issues. Besides the smaller numbers of nuclei in the gametophytes (i.e. lowering precision of the analysis) such a small shift might be caused by different levels of cytosolic compounds between the gametophytes and mature (sporophyte) leaves, which can influence fluorescence staining (Doležel et al., 2007).

Sporogenesis

Sporogenesis of apomictic Dryopteris affinis agg., including pentaploid D. × critica, was comprehensively studied by Manton (1950) (note that all cytotypes are marked as D. borreri in that study) and Schneller (1975) (under the names D. pseudomas and D. × tavelii). Whereas in apomictic diploids and triploids the eight-celled type of sporangium prevails, in tetraploid and pentaploid hybrids of an apomictic and sexual species, the 16-celled type predominates. The sporangia of hybrids are exceptional and combine the normal apomictic development with aborted and well-developed spores. Manton (1950) proposed that in these plants, there are a few large good spores that germinate and produce gametophytes and consequently sporophytes. These plants (F₂ generation) were not successfully analysed by Manton (1950) and died in culture. We repeated Manton’s experiment and surprisingly revealed two size classes of spores and the 2·5x (reduced) plants and 5x (unreduced) plants arisen from the F₁ pentaploid hybrid. The existence of viable plants originating from reduced spores of the odd-ploidy-level parent has never before been observed in ferns.

Manton (1950) and Schneller (1975) also examined sporogenesis of the same (or similar) hybrids in the Dryopteris affinis agg. They described the predominant formation of 16-celled (sexuality-like forming 64 spores) together with eight-celled (apomictic type forming 32 spores) sporangia on one plant. Manton (1950) also discovered an ‘intermediate’ type of sporangium and described its sporogenesis in detail in Dryopteris borreri s.l., D. atrata, D. remota and Pteris cretica. She observed that one or several restitution nuclei in an apomictic-type sporangium may exhibit irregularities leading to division into two unequal parts. Meiosis is then regular even in small nuclei, but due to unbalanced numbers of chromosomes the spores abort (Manton, 1950, p. 166). Because not all restitution nuclei are involved in this process, the resultant number of spores higher than 32 (no irregular division) and lower than 64 (division of all nuclei) and a mixture of unbalanced aborted spores together with ‘normal’ diplospores within a sporangium may be expected. These counts were usually studied in immature sporangia. Further deviation from the standard apomictic/sexual pathway was documented by Schneller (1975) who observed not only sporangia with either eight or 16 spore mother cells but also sporangia with intermediate counts and unequal cell size. This may indicate that some spore mother cells in a sporangium underwent the last mitosis (are reduced) while others did not (unreduced). Moreover, Schneller (1975) also reported the extremely rare occurrence of sporangia with only four spore mother cells that had probably twice as many chromosomes as the maternal plant due to failure of two mitotic divisions.

Hitherto, studies focused on the number of spores per sporangium reported either 32 spores per sporangium for apomictic species or 64 for sexual species without exceptions (e.g. Gastony and Haufler, 1976; Regalado Gabancho et al., 2010; Huang et al., 2011; Dyer et al., 2012). The first indication of an unbalanced spore number in a single sporangium was recently presented in the peculiar case of diploid sexual Phegopteris decursive-pinnata; the variation was caused probably by rare mutations disturbing meiosis (Nakato et al., 2012). Our study for the first time suggests the presence of different (apomictic vs. sexual) modes of spore mother cell development in a single sporangium. Formation of only aborted spores or a mixture of aborted and either type or both types of well-developed spores (the two size classes, probably corresponding to reduced and unreduced spores) in a single sporangium was recorded (Table 2; Fig. 3). The number of spores was between 32 (expected for full apomixis) and 64 (full sexuality). Together, these facts show that all the different processes are included not only in one sorus but even in one sporangium. It seems that spore mother cells are more or less independent and each can develop into different types of spores, and moreover that irregularities described first by Manton (1950) (see above) may co-occur. The numbers of well-developed spores per sporangium were multiples of four or close to it (Table 2, this study). This suggests that when reduced and/or unreduced well-developed spores occur, they are produced in whole tetrads. Deviations from exact multiples of four might be caused by occasional abortion of some spores and/or counting errors (especially in case of one ‘excessive’ spore, as the distinction between aborted and well-developed spores is not always clear-cut and an error of ±1 could occur).

The formation of unreduced diplospores in apomictic fern hybrids that give rise to new sporophytes has been reported by several studies (e.g. Walker, 1984; Rabe and Haufler, 1992; Chao et al., 2012). In contrast, evidence for a mixture of reduced and unreduced spores on one plant is very sparse. The first evidence was provided by Hickok and Klekowski (1973) in Ceratopteris hybrids. Their study indicated the presence of meiotic adaptations within hybrid sporophytes that allow for the production of viable unreduced spores and gametophytes as well as reduced spores. Dyer et al. (2012) reported the occurrence of presumably reduced (64 spores per sporangium) and unreduced (32 per sporangium) spores on one individual of apomictic Asplenium hallbergii. Joint production of aborted, reduced and unreduced spores in different plants of the same population of Phegopteris decursive-pinnata was recently studied by Nakato et al. (2012).

Evolutionary implications

Our study revealed an unusual pattern of ploidy levels among F₂ offspring of pentaploid Dryopteris × critica. The finding that pentaploid hybrids can produce new viable plants of reduced 2·5x ploidy level, or even reduced and unreduced offspring on one plant, is particularly important for understanding the possibility of ploidy level reduction in ferns. Production of viable reduced spores, instead of aborted or unreduced diplospores, has important consequences only for hybrid formation. This contrasts with the general expectation that apomixis is likely to be established in triploid and pentaploid hybrids to avoid their sterility (Liu et al., 2012). In particular, the formation of reduced spores from odd-ploidy hybrids arising from sexual and apomictic species can be an important mechanism for the formation of new entities (see also similar cases by Rabe and Haufler, 1992; Nakato et al., 2012) and possible diploidization in polyploid ferns. The existence of meiosis and consecutive ploidy reduction in a hybrid polyploid entity may play an important and yet undetected role in fern speciation.

In Fig. 5, we present the most likely hybridization schemes of the sexual and apomictic fern species, considering the formation of polyhaploids. Lovis (1977) speculated that most apomictic ferns are triploids (50–70 %) or diploids (20–35 %). In two cases (B and C, hybridization of an apomictic diploid and a sexual tetraploid and of an apomictic triploid and a sexual diploid), the resulting hybrids are tetraploid, which allows formation of diploid reduced offspring of the new genetic composition. It can be expected that these reduced diploids are genetically stable due to even numbers of chromosome sets, especially in case B, in which the tetraploid hybrid has two chromosome sets from each parent. The other two cases (A and D) lead to odd-ploidy hybrids resulting in possibly unstable aneuploid polyhaploids (D being the case of D. × critica in the present study).

Fig. 5.

Theoretical scheme of the most likely crosses between sexual and apomictic fern species. Formation of hypothetical polyhaploids is considered.

Although we have not found polyhaploids (2·5x plants) in natural populations, we believe they might have certain evolutionary potential. There are several reasons that make detection of polyhaploids in wild populations difficult. In our case, the pentaploid F₁ hybrids are rare in most of the studied populations (approx. 10 % of D. affinis s.l. plants) and only a small part (approx. 10 %) of their spores are not aborted (compared with most of the viable spores in parental species). Moreover, only a small proportion of hybrid F₂ gametophytes produced sporophytes (approx. 10 %). Combined together, these three frequencies determine that the overall frequency of polyhaploids is much below 1 %, even if we assume the same fitness of all types of gametophytes and sporophytes (which might not be the case). However, in small populations, some of these frequencies (especially the frequency of hybrid plants) might be enhanced, resulting in a more significant frequency of polyhaploids. On the other hand, such populations are difficult to find in the field and sampling the representative number of individuals (i.e. finding and analysing many such populations) is nearly impossible. We should also consider that D. × critica has an odd ploidy level, which leads to chromosomally unstable polyhaploids. In the case of tetraploid hybrids and especially the case shown in Fig. 5B, more regular formation of polyhaploids (fewer aborted spores, higher rate of sporophyte formation) can be expected, leading to higher polyhaploid frequencies.

CONCLUSIONS AND FUTURE RESEARCH

Our study has demonstrated the occurrence of the mixed reproductive mode in an apomictic × sexual fern hybrid. Two types of functionally viable spores are produced: unreduced (apomictic) 5x diplospores and reduced (sexual) 2·5x spores. The existence of reduced viable spores and the occurrence of both types on one plant and even in one sporangium together is unexpected and novel. Moreover, both spore types are capable of successful sporophyte production, which has not previously been observed. The pentaploid hybrid is capable of autonomous reproduction. In general, the apomictic × sexual hybrids might be of certain evolutionary potential, particularly if their polyhaploid offspring are capable of producing viable spores and crossing with sexual species. To investigate this, we will continue cultivation of polyhaploids until they reach maturity. Many other interesting research topics are raised based on our data, such as the incidence of polyhaploid formation in other fern groups or ploidies, genetic variation of polyhaploid offspring, and especially the occurrence and fertility of polyhaploids descendant from tetraploid hybrids (the scheme in Fig. 5B).