Flow cytometric sorting of progeny origins and ploidy level

The seeds that were sampled from open-pollinated Pilosella plants in the field were analysed for way of origin using Flow Cytometric Seed Screening (FCSS; Matzk & al. 2000). The method was modified for Pilosella as follows. Fresh seeds (achenes) collected from fruiting plants were either analysed shortly after harvest or they were kept in a fridge. Cold storage preserves the endosperm tissue, allowing interpretable endosperm peaks in flow histograms up to several months after collection. Before the analysis, the filled (well-developed) seeds were screened using a stereomicroscope and an adequate pressure of tweezers on each seed. The filled seeds were pooled either two at a time or ten at a time (plus the rest of at least six seeds), so either the seed doublets or the groups comprising usually ten seeds each were prepared for flow cytometric analysis. The DAPI staining method (Otto 1990) was used, using the ice-cold nuclei-extracting buffer (Otto I) and the staining buffer (Otto II) supplemented by 2 µl ml^-1 mercaptoethanol as an antioxidant. The fluorescence intensity was determined by a Ploidy Analyser PA II (Partec GmbH, Münster, Germany) equipped with an HBO high-pressure mercury lamp for UV excitation. If possible, 5000 nuclei were scored in each sample.

Analysis of seed doublets (Krahulcová & al. 2009a, 2011) — Two achenes were chopped in nuclei-isolating buffer with a razor blade together with 0.05 cm² of leaf tissue of reference standard (Bellis perennis, diploid Pilosella lactucella or some of the polyploid Pilosella species: the standard was chosen according to ploidy level of the maternal plant of analysed seed progeny). The suspension was filtered and incubated for 10 min at room temperature, stained and analysed. The histograms with CVs below 3.5% were considered.

Analysis of pooled samples comprising six to ten seeds (Krahulcová & Suda 2006; Krahulcová & al. 2011) — Pooled achenes were chopped in nuclei-isolating buffer with a razor blade, the suspension was filtered and incubated for 10 min at room temperature, stained and analysed. The histograms with CVs below 5% were considered. This threshold of accuracy of analyses was still acceptable because two simultaneous peaks in a histogram should distinguish the difference in DNA content among components within a sample that corresponds to double CV (Doležel & Göhde 1995). Quantification of the embryo ploidy classes was based on the proportion of nuclei in their respective embryo ploidy peaks, and the maternal ploidy peak was used as an internal ploidy standard. In most histograms the minor endosperm peak was evident, which was positioned either at 2× the distance of the embryo peak assigned to embryos sharing the ploidy of respective polyploid apomictic mothers, or at 1.5× the distance of the embryo peak assigned to embryos sharing the ploidy level of sexual mothers.

The FCSS procedures allowed detection of both ploidy level in progeny embryos (when analysing both the seed doublets and pooled samples of a maximum of ten seeds) and their reproductive origins (when analysing the seed doublets). The FCSS analyses of pooled samples comprising six to ten seeds revealed the most frequent embryos maintaining the maternal ploidy level; in addition to them, the analyses could distinguish and quantify the potential less frequent progeny embryos that had a ploidy level different from a maternal parent, namely, the products of (1) heteroploid crosses, (2) sexual mating via unreduced gametes and (3) haploid parthenogenesis (Krahulcová & Suda 2006). The seed-progenies, produced by both facultative apomicts and sexuals, were scored. Analysing the seed doublets, we distinguished the progeny generated by apospory versus sexual mating; analysing the pooled samples of six to ten seeds, we traced a potential diversity in embryo ploidy level (Krahulcová & Suda 2006; Krahulcová & al. 2009a, 2012, 2014). The results of these two procedures of seed screening were evaluated separately. Consequently, to illustrate a residual sexuality in facultative apomicts, the “off-type” progeny that did not replicate the maternal parent (Bicknell & Koltunow 2004) were detected and quantified. The capacity of facultative apomicts to avoid apomeiosis by forming a reduced female gamete was tested using emasculation experiments (Krahulcová & al. 2004). The seeds sampled from emasculated apomictic plants were also analysed using the FCSS analysis of pooled samples comprising six to ten seeds. This way, the embryos maintaining the maternal ploidy level (those generated from unreduced female gametes via apospory) were distinguished from embryos having a half of maternal ploidy level (those generated from reduced female gametes via haploid parthenogenesis).

Crossing experiments

The capacity of selected pairs of parental accessions to hybridize was tested using hand crossing experiments. A restricted extent of crossing experiments was caused by availability of parental accessions in the experimental garden that flowered simultaneously in sufficient replicates. Namely, three hexaploid facultatively apomictic accessions of Pilosella bauhini were chosen as maternal parents and crossed with a sexual (mostly diploid) pollen parent. A high pollen stainability detected in pollinating species/ cytotypes suggested a sufficient pollen viability of pollen donors (Rotreklová & Krahulcová 2016). The diploid sexual P. ^* testimonialis was used as a maternal parent exclusively in the reciprocal cross (Table 2). All inflorescences (capitula) of each parent were isolated in nylon bags before anthesis (prior to the outer ligular flowers of the capitulum opening). The capitula were crossed in the stage of stigma receptivity (when bifurcate stigmas protruded from the flowers), by rubbing the whole parental capitula together. As the flowers in the capitulum open successively from the margins to the centre, the maternal capitulum was usually pollinated 2 to 3 times a week, on each occasion with a new capitulum of the same pollen plant. The pollinated capitula were kept in nylon bags until the seeds ripened (Krahulcová & al. 2004).

The seed-progeny gained from crosses was evaluated using the FCSS method in the same manner used for those progeny generated by open-pollinated plants in the field. Because the FCSS method destroys the seed, a portion of seeds gained from crossed maternal plants with an abundant seed-set was used for seedling recovery. The ploidy level of these cultivated seedlings was determined using flow cytometry with the DAPI staining method (Krahulcová & al. 2004). Approximately 0.3 cm² of fresh leaf tissue from the seedling was chopped together with the same amount of leaf tissue from the maternal plant (internal standard) in nuclei-extracting buffer, then the same method used for analysis of seeds (described above) was followed.

Fig. 2.

Schematic drawing illustrating putative spontaneous crosses among Pilosella species in Bulgaria. Selection of taxa is based on systematic field studies by Krahulcová & al. (2009b, 2016). Except for P. cymosa subsp. sabina (shown as “P. ^*sabina”) and P. rhodopea, the parental basic species co-occurred with respective recent hybrids in 19 populations examined (Table 1). Position on circle does not exactly reflect relationships among individual taxa. Number of hybridizing populations corresponds to entries in Table 1.

Attributes and distribution of species occurring in both regions

Present studies regarding recently hybridizing populations of Pilosella cover three parental basic species that occur in both Central Europe and Bulgaria, namely, P. bauhini, P. officinarum and P. onegensis (Fig. 2, Fig. 4). The first two species, which vary in ploidy level, are widespread in both regions. However, P. officinarum exhibits a specific pattern of cytotype variation in Central and SE Europe (Mráz & al. 2008). The diploid sexual P. onegensis is rather common in Bulgaria but rare in Central Europe. Despite this difference, recent hybrids of this species were only sporadically found in both regions (Fig. 2, Fig. 4). Facultatively apomictic biotypes of P. bauhini (both pentaploid and hexaploid) form interspecific hybrids in both regions in a similar manner. However, P. bauhini exhibits major ploidy variation in Central Europe, ranging from tetraploid to heptaploid ploidy levels. The tetraploids are predominantly sexual and less frequently apomictic, whereas the other cytotypes are apomictic (Rotreklová 2004; Krahulcová & al. 2009a, 2014). Interestingly, the pentaploid and hexaploid apomictic biotypes of P. bauhini exhibit substantial genotypic diversity in Central Europe both within and among populations (Krahulcová & al. 2009a, 2014 and unpublished data; Rosenbaumová & Krahulec 2015).

The hybrids of two basic species mentioned above, Pilosella bauhini and P. officinarum, are common both in Bulgaria and Central Europe (Fig. 2, Fig. 4). The species P. officinarum generates interspecific hybrids with other co-occurring species in both regions (Fig. 2, Fig. 4). Nevertheless, the biotypes of P. officinarum that commonly hybridize in Bulgaria differ from those in Central Europe, primarily with respect to ploidy level and reproduction mode. Namely, P. officinarum, which was frequently recorded as a putative parental species in hybridizing populations in Central Europe, is tetraploid and sexual (Fig. 4). This biotype of P. officinarum is absent in Bulgaria and is substituted by facultatively apomictic pentaploids and hexaploids that are considered to be putative parents of interspecific hybrids in this region (Fig. 2; Krahulcová & al. 2009b, 2016). Function of the sexual tetraploid P. officinarum in shaping of the genus Pilosella in Central Europe may be comparable to the role of sexual tetraploids in Rubus subg. Rubus (Sochor & al. 2015; Šarhanová & al. 2017). These taxa with a high level of sexuality that are represented by both R. ser. Glandulosi (Wimm. & Grab.) Focke and an obligately sexual R. caesius L., force the evolution and spéciation of R. subg. Rubus in Central Europe (Sochor & al. 2015; Šarhanová & al. 2017).

Capacity for sexual 2n + n mating in apomicts

A strongly different capacity for sexual mating via unreduced gametes predominantly from an apomictic parent was confirmed in Bulgaria and Central Europe. The seed progeny that originated in the field in this manner was recorded in six out of eight mixed apomictic-sexual populations in Central Europe (Krahulcová & al. 2014; Rosenbaumová & Krahulec 2015), reaching the maximum frequency of 7% out of all progeny embryos that were formed by apomictic mothers in a population (Fig. 3 in Krahulcová & al. 2014). By contrast, no such progenies were recorded in nine analogous populations studied in Bulgaria (Fig. 3A, 3B).

The experimental crosses simulating the spontaneous hybridizations in the field also provided specific results in both regions. Two polyploid facultatively apomictic species from Central Europe, Pilosella bauhini and P. aurantiaca, were crossed with the sexual counterpart P. officinarum from this region. When either the pentaploid P. bauhini or the tetraploid P. aurantiaca were pollinated by the tetraploid P. officinarum, no polyhaploid (n + 0) progeny seedlings were formed, and a substantial part of the sexually derived progeny was of 2n + n origin (Table 2 in Krahulcová & al. 2014). On the other hand, the off-type progeny seedlings recovered from a cross of the hexaploid P. bauhini from Central Europe and another tetraploid clone of P. officinarum comprised the prevailing n + n hybrids as well as the rarer trihaploids (n + 0) and the hybrids that originated via an unreduced gamete of either parent (2n + n, n + 2n) (Rosenbaumová & Krahulec 2015). Parallel crosses of the apomictic hexaploid P. bauhini from Bulgaria and a sexual parent yielded various proportions of the offtype progeny, probably reflecting the taxonomic/ploidal affinity of the parents (Table 2). Nevertheless, sexual mating via unreduced gamete(s) was rarely recorded, and a substantial part of the off-type progeny included trihaploids of n + 0 origin (Table 2). Consequently, the specific parental ploidy relations that are attributed to the two compared regions might primarily influence the structure of progenies potentially generated from interspecific crosses.

Composition of biotypes affects specific microevolution in mixed populations in both regions

All putative hybrids sampled in the field in Bulgaria are considered recent because they were not spread outside the hybridizing populations and mostly coexisted with both putative parents as individual plants (Table 1). Unlike in Central Europe, no widespread stabilized hybridogeneous taxa that occur independently of their parental species were found in Bulgaria except Pilosella guthnikiana (Hegetschw.) Soják (Vladimirov & Szelag 2001). The presence of hybridogeneous species, both of n + n and 2n + n origin (e.g. Fehrer & al. 2007), leads to a more complicated structure of Pilosella populations in Central Europe in comparison with Bulgaria. In Central Europe, the hybridogeneous species behave independently of parents and often hybridize with other co-occurring taxa thereby forming new combinations. It is not rare to find there multiple hybrids where the morphological characters indicate an involvement of three or rarely four parental basic species (e.g. Křišt'álová & al. 2010; Bräutigam 2017; and Fig. 4 in this paper). By contrast, we did not record any such complex multiple hybrids during our six-year population studies in Bulgaria.

The fundamental difference in both areas involves the cytotype structure of (potentially) hybridizing taxa: (1) whereas the sexual diploid and apomictic polyploid taxa often co-occur in hybridizing populations in Bulgaria (Table 1), the diploid sexual putative parents were rather rarely recorded in hybrid swarms in Central Europe (Table 3). (2) Unlike in Bulgaria, sexual taxa exceeding the tetraploid ploidy level occur in Central Europe (Rotreklová & al. 2002; Mráz & al. 2008; Trávníček & al. 2011). Similarly, the tetraploid sexual biotype of Pilosella officinarum, which is the most common sexual parent of both recent (Table 3) and stabilized hybrids in Central Europe (e.g. Suda & al. 2007; Bräutigam 2017), is absent in Bulgaria. (3) In addition to P. officinarum, numerous other tetraploid hybridizing taxa occur in Central Europe, whereas tetraploids are rather rare among Pilosella species in Bulgaria. Therefore, the interspecific crosses occurring in the Central European populations often involve at least one tetraploid parent (Table 3). Consequently, the difference in parental ploidy level is reduced in most of the hybridizing species in Central Europe (Table 3) compared with Bulgaria (Table 1), suggesting the impact of stronger inter-ploidy hybridization barriers in Bulgaria compared with Central Europe.

Many recent hybrids established in the Central European populations were formed via an unreduced gamete. Six out of 14 examples of such recent 2n + n hybrids (Table 3) originated via fertilization of an unreduced gamete from an apomictic maternal parent (2n + n hybrids), whereas the sexual parent formed an unreduced gamete most likely in two crosses only (Table 3). The determination of the maternal lineage in respective hybrids in Central Europe was possible due to a specific haplotype shared by the maternal parent and the hybrid (Table 3). The apomictic 2n + n hybrids that repeatedly originate from an apomictic and a sexual parent and exceed the parental ploidy level play an important role as sources of population diversity in Central Europe (summarized in Krahulcová & al. 2014). Although their genome is unstable due to a high level of residual sexuality, such hybrids stimulate the origin of entirely new hybridogeneous types because they produce an extremely variable progeny. On the other hand, in 19 Bulgarian populations, we detected only a single recent hexaploid hybrid Pilosella bauhini × P. ^*testimonialis (row 15 in Table 1) that might have unambiguously originated from a simple cross via fertilization of an unreduced gamete from an apomictic pentaploid parent P. bauhini.

Table 3.

Pilosella species (sexual species underlined) and their putative recent interspecific hybrids recorded at 18 localities in Central Europe (for data referring to taxonomy, ploidy level and description of localities, see references below table). Hybrid status was inferred from a combination of morphology and ploidy level of both putative parental species that co-occurred with the hybrids at the respective locality. Bold cytotypes refer to 14 simple (primary) hybrids that originated via an unreduced gamete from either parent (2n + n, n + 2n). Out of them, only two hybrids () might originate via an unreduced gamete from a sexual parent. The 18 hybrid cytotypes that shared the maternal cp-haplotype () originated verifiably via fertilization of the maternal facultatively apomictic parent, suggesting either an unreduced (six hybrids) or a reduced female gamete (12 hybrids). Cytotypes considered as multiple hybrids either had a ploidy that cannot be inferred from a simple hybridization of co-occurring parental cytotypes, or their morphology was markedly closer to either parent, suggesting a backcross. Seedsterile/ semisterile hybrids are indicated „!”. Abbreviations: CZ = Czech Republic; D = Germany. Parental species comprise five basic species (P. aurantiaca, P. bauhini, P. lactucella, P. officinarum, P. onegensis) and four intermediate species (P. floribunda, P. glomerata, P. polymastix, P. rubra).

The polyhaploid (n + 0) seed progeny generated at various rates by facultatively apomictic polyploids was detected in populations in the field in Bulgaria (Fig. 3A, 3B) and Central Europe (Krahulec & al. 2011). However, reliable detection of the polyhaploid origin is possible only in those progeny embryos and seedlings that can be compared with their certainly identifiable maternal parent (e.g. Krahulec & al. 2008). When examining the mature potentially polyhaploid plants already established in the field, this requirement is unfeasible. Nevertheless, mature triploid plants that were likely of polyhaploid origin were rarely found in two populations in the Czech Republic (Krahulec & al. 2008, 2014). This finding admits the importance of polyhaploids in microevolutionary processes that were simulated in experiments (Krahulec & al. 2011). Because no triploids that might potentially be of polyhaploid origin (n + 0) were detected among mature plants in populations in Bulgaria (Table 1), the polyhaploid progeny likely exhibits a minimal chance to establish and survive under field conditions there. Consequently, the haploid parthenogenesis that along with a sexual mating generates progeny not replicating the apomictic parent appears to be unsuccessful in populations in Bulgaria.

Evidence and impact of hybridization in other apomictic genera – examples in Asteraceae and Rosaceae

Hybridization has played an important role in evolutionary history of Hieracium s. str. that is closely related to Pilosella. Both genera, formerly classified as subgenera of the genus Hieracium s.l., share a traditional classification that is based on morphology and distinguishes the basic and intermediate (hybridogeneous) species. Surprisingly, using a combination of molecular markers, many of the species of Hieracium s. str. (diploid to tetraploid) that had traditionally been classified as basic ones, proved to be of hybrid origin (Krak & al. 2013). This suggests a stronger effect of hybridization on evolution of the genus that had been assumed previously. On the other hand, unlike to Pilosella, the recent Hieracium hybrids potentially originating in the species-mixed populations are very rare: using the combination of nuclear DNA markers and allozymes, one recent hybridization between two diploid species was still documented (Mráz & al. 2011).

Another agamic complex with a complicated reticulate structure, Taraxacum F. H. Wigg. (dandelions, Asteraceae), also comprises sexual diploids and apomictic polyploids similarly to hawkweeds (both Hieracium s. str. and Pilosella). However, unlike in Pilosella, apomixis is considered to be almost obligate in Taraxacum, with the exception of an occasionally fertilized unreduced female gamete (Mártonfiová & al. 2010). Therefore, a rare manifestation of residual sexuality in apomictic dandelions limits a potential impact on interspecific hybridization. The asexual clones of Taraxacum that had probably originated primarily from crosses between sexuai\ls and apomicts have further been diversified by means of accumulation of mutations (Majeský & al. 2012). In fact, a perfect correlation was found between genotype fingerprints and morphotypes in polyploid obligatory apomictic accessions of T. officinale agg. (Majeský & al. 2012), and the clusters of apomictic taxa in T. sect. Erythrosperma (H. Lindb.) Dahlst. may also be well characterized by both molecular markers and morphology (Majeský & al. 2015).

A various congruence of molecular markers and morphological characters delimiting the taxonomic groups at different levels was found in the family Rosaceae. Whereas the standard barcoding markers resolved the taxonomic groups in Crataegus L. poorly due to supposed hybridization between different lineages of the genus (Zarrei & al. 2015), a genetic evidence of parentage in 11 hybrid species of Sorbus L. in Central Europe was given, combining the chloroplast DNA and AFLP markers (Uhrinová & al. 2017). These Sorbus hybrids among sexual parents are polyploid and supposedly apomictic and may or may not be discriminated by morphology (Uhrinová & al. 2017). The combination of nuclear and chloroplast genetic markers was speciesspecific, distinguishing all of the few ancestral diploid species of Rubus subg. Rubus (Sochor & al. 2015). Furthermore, hybridogeneous origins and parentage were identified in several polyploid and apomictic units (series) of this subgenus that conventionally join the species sharing a similar morphology: these analysed polyploid Rubus taxa above species level are believed to be stabilized hybrids of ancient origin (Sochor & al. 2015). Furthermore, a genetic evidence of hybrid origin was accomplished in both stabilized apomictic species and local recent hybrids comprising one of the series of R. subg. Rubus (Šarhanová & al. 2017).

The pattern of reproductive variation and the processes that putatively shape population structures in Pilosella and Rubus subg. Rubus share some similar features. The level of apomixis/sexuality in the tetraploid R. ser. Glandulosi proved to be also different in two Central European regions: the same taxa were strictly sexual and partly apomictic, respectively (Šarhanová & al. 2012). Similarly to Pilosella in Central Europe, the embryos of 2n + n origin increase the ploidy level in the seedprogenies of R. subg. Rubus in the field (Šarhanová & al. 2012). These embryos that had originated via fertilization of an unreduced female gamete of both triploid and tetraploid mother plants were detected in three series of Rubus with a high degree of residual sexuality. In this way, the origin of the pentaploid and hexaploid species of R. subg. Rubus is suggested (Šarhanová & al. 2012). On the other hand, this type of rare sexual mating is probably the only opportunity for the manifestation of residual sexuality in apomictic mothers in Rubus, because no sexual n + n progeny embryos were detected in apomictic taxa (Šarhanová & al. 2012). The hybridization in R. subg. Rubus is thus almost exclusively unidirectional (Šarhanová & al. 2017), similarly to hybridization in apomictic taxa of Taraxacum (Mártonfiová & al. 2010).