Newly unveiled meiosis elucidates the unreduced gamete frequency and its impact on the evolution of the Lemna minor complex Authors: Yuri Lee (이유리), Veit Schubert, Anton Stepanenko, Gihwan Kim, Luca Braglia, Ingo Schubert, Laura Morello Published: 2026 ABSTRACT AbstractFusion of gametes possessing meiotically reduced (haploid) chromosome complements is the main pathway of propagation among eukaryotes. However, duckweeds, the smallest angiosperms, propagate mainly vegetatively, and meiosis has not yet been documented in detail for this plant family. More surprising was the recent evidence of rather frequent interspecific hybrids and triploid clonal accessions, which became obvious by genome size measurements, genomic in situ hybridization (GISH), and combined plastid and nuclear DNA markers. These observations indicated sexual propagation involving reduced as well as unreduced male and female gametes in Lemna minor and L. turionifera, leading to allodiploid and allotriploid hybrids (with genome compositions MT, MMT, MTT) and autotriploid L. minor (MMM) accessions. Here, we (i) document the meiotic stages of Lemna species for the first time photographically; (ii) provide evidence of unreduced male gametes through fluorescent in situ hybridization (FISH) with single locus probes; (iii) determine their abundance in different individuals; and (iv) discuss possible reasons for unreduced male gamete formation. These findings open new insights into the modes of sexual reproduction and evolution of duckweeds, which may be useful for future breeding efforts in this emerging crop. Introduction Meiosis, as a reductive cell division, is essential for sexual reproduction, including recombination of paternal and maternal alleles. It comprises a single round of DNA replication followed by two cell divisions, separating homologous chromosomes during the first, and sister chromatids during the second division, thus generating cells with half the number of chromosomes of the somatic mother cell. However, meiotic defects can produce viable gametes retaining somatic chromosome numbers, potentially leading to polyploid progeny (Brownfield and Köhler, 2011; Zamariola et al., 2014). Plant polyploidy is widespread in natural ecosystems and in many major crops (Brownfield and Köhler, 2011). Therefore, understanding the cytological events underlying normal meiosis as well as the formation of unreduced gametes is fundamental for elucidation of evolutionary processes, and provides a basis for plant breeding. Duckweeds, the smallest flowering angiosperms, are an aquatic monocot family (Lemnaceae Martinov) which mainly reproduce through fast asexual propagation. Lemna L., the largest of five duckweed genera, comprises four sections (Alatae Hegelm., Biformes Landolt, Lemna L., and Uninerves Hegelm.), including so far 11 species and three interspecific hybrids, as well as ploidy variants and backcrosses (for review see Morello et al., 2026). Even though duckweeds have been studied for over two centuries, due to their small size (0.6–9 mm in diameter), simple organismic structure (leaf-like fronds with or without roots), and rare flowering under natural and/or experimental conditions, their taxonomy and phylogeny remained unclear until recently. Recent applications of advanced molecular and cytological approaches have revealed surprisingly frequent auto- and allopolyploids, which require sexual propagation (Braglia et al., 2024; Ernst et al., 2025; Michael et al., 2026; Stepanenko et al., 2025). Notably, the representative interspecific hybrid in sect. Lemna, Lemna × japonica Landolt, results from unidirectional crosses between L. minor L. (M) ♀ and L. turionifera Landolt (T) ♂, and exhibits variation in ploidy and genome composition (MT, MMT, MTT) (Braglia et al., 2021; Ernst et al., 2025), implying recurrent hybridization and the occurrence of unreduced gametes in both parental species. Consistent with this inference, naturally occurring autotriploid L. minor accessions have also been reported (Michael et al., 2026). However, direct experimental confirmation of unreduced gamete formation has been lacking so far. In most flowering plants, interploidy crosses via fusion of reduced and unreduced gametes usually result in abnormal endosperm development and seed abortion, called ‘triploid block’ (Cooper and Brink, 1945; Marks, 1966; Köhler et al., 2021). Duckweed can overcome this reproductive barrier, at least in part, possibly through the loss of genes of RNA-dependent DNA methylation (RdDM) pathways (Ernst et al., 2025). Their predominant clonal growth facilitates the persistence of triploid lineages and their sympatric diversification, despite sterility (Lee et al., 2026). Consequently, knowledge about the frequency of reduced and unreduced gametes in duckweeds is instrumental for elucidating their evolutionary dynamics. Due to many small chromosomes and only a few meristematic cells, reliable cytological studies are scarce for duckweeds (e.g. Cao et al., 2016; Hoang and Schubert, 2017; Hoang et al., 2019, 2020, 2022; Stepanenko et al., 2025). Moreover, the rarity of natural flowering, the difficulty of inducing flower formation under experimental conditions, and the smallness of flowers (flower <1 mm with one pistil and asynchronously growing two stamens surrounded by a spathe; see Supplementary Fig. S1 in Lee et al., 2026) prevented detailed studies of meiosis, despite the obvious need for assessing the evolutionary prospects of duckweeds. We are aware of only a very limited number of studies mentioning duckweed meiosis (Caldwell, 1899; Maheshwari and Kapil, 1963) without photo-documentation. This study provides the first detailed cytological characterization of male meiotic stages, a protocol for their preparation from young flowers, and for analysing the diverse final meiotic products in the duckweed section Lemna. By examining L. minor and L. turionifera, we quantified the meiotic products of different ploidy levels, and assessed the occurrence and frequency of reduced and unreduced gametes in these predominantly clonally propagating plants. Materials and methods Plant materials and culture conditions All tested duckweed accessions were originally from the Landolt collection and are now available at the duckweed collection of the Institute of Agricultural Biology and Biotechnology (CNR-IBBA) in Milan (Italy; Morello et al., 2025). The diploid accessions L. minor 5500, L. turionifera 9434 and 7432 were selected based on prior investigations of their floral development, as well as their genetic and cytogenetic features (Braglia et al., 2021; Ernst et al., 2025; Lee et al., 2026). Plants were maintained under axenic conditions (see Braglia et al., 2021). Preparation of meiotic cells To prepare meiotic stages, young developing anthers were collected from artificially induced flowers (Lee et al., 2026). In brief, few fronds were pre-cultured for seven days in 50 ml plastic beakers with 20 ml of SH medium (Schenk and Hildebrandt, 1969), including 1% sucrose. For flower induction (FI), some asexually propagating, healthy fronds were then placed in 100 ml plastic beakers with 100 ml of modified NH4+-free 5% Hutner liquid medium (Lee et al., 2026) and 30 µM salicylic acid (SA). The plants were kept under a 16 h light/8 h dark photoperiod (white light, 150 μmol m‒2 s‒1) at 25 °C. Considering that Lemna plants usually start flowering from days 12–14 of FI (Lee et al., 2026), at day 9, fronds were transferred into 90×20 mm petri dishes containing semisolid FI medium (with 0.3% agar), to monitor for expected immature flowers (possibly undergoing meiosis). Fronds bearing young flowers were confirmed by checking the underside of the petri dish under a stereo microscope (SMA18, Nikon, Tokyo, Japan) (Fig. 1). Immature flowers of different sizes were independently dissected and immediately fixed in a freshly prepared 3:1 (v:v) ethanol:acetic acid solution for 3–6 d at 4 °C, then transferred to 70% ethanol and stored at −20 °C until further use. Different preparation stages of one individual of Lemna turionifera clone 7432 bearing immature flower. (A) Abaxial side of three cohering fronds, one of which is flowering; (B) the detached flowering (fl) frond; (C) the young flower surrounded by the reddish spathe (sp) is visible after removing frond’s abaxial surface; (D) the young, developing primary stamen (st1) and the secondary stamen (st2) are visible after removing the spathe. Scale bars=500µm. Prior to slide preparation, fixed flowers were treated with 45% acetic acid for up to 30 min. Single anthers were transferred onto glass slides with 1 µl of 45% acetic acid, and further dissected using hypodermic needles (18G) to release the meiotic cells. After adding 7 µl of 45% acetic acid, the cell suspension was covered with a coverslip. Slides were briefly flamed over a burning ethanol burner (∼1 s), then immersed in liquid nitrogen for 10 s. The coverslip was subsequently removed using a razor blade, and the slides were immediately placed in pre-chilled 99.8% ethanol at −20 °C. Probe generation and fluorescence in situ hybridization (FISH) Lemna-specific 5S and 45S rDNA probes were generated following the procedure described by Hoang and Schubert (2017) and Stepanenko et al. (2026) with small modifications. Total DNA was extracted from plant tissue using a modified CTAB method (Murray and Thompson, 1980). First, genomic DNA isolated from L. minor (clone 8703) was amplified using duckweed rDNA-specific primers (Hoang and Schubert, 2017; Stepanenko et al., 2026). After purification, the obtained PCR products were used as templates for labeling with specific fluorescent dyes. The rDNA fragments were subjected to nick-translation using the AZDye594 NT Labeling Kit for 45S rDNA (red) and the AZDye488 NT Labeling Kit for 5S rDNA (green) (Jena Bioscience, Jena, Germany). After precipitation with 96% ethanol, probe pellets were dissolved in 100 μl hybridization buffer (50% (v/v) formamide, 20% (w/v) dextran sulfate in 2×SSC, pH 7). The meiotic products prepared on slides were denatured in 2N NaOH for 5 min at room temperature (22∼25 °C), rinsed for stepwise dehydration in 75% (twice) and 99.8% ethanol for 5 min each, and air-dried. Of each FISH probe, 1 µl (50 ng μl‒1) was added to 10 μl hybridization mixture [dextran sulphate (1 g) dissolved in 1 ml ddH2O, followed by addition of 5 ml deionized formamide, 1 ml 20×SSC, and 1 ml sonicated fish sperm DNA], denatured at 99 °C for 15 min, and stored at −20 °C until use. The mixture was applied to air-dried slides, sealed with coverslips, and incubated in a moist chamber at 37 °C overnight (Aliyeva-Schnorr et al., 2015). After hybridization, the slides were rinsed twice in 2×SSC for 5 min, followed by distilled water for 2 min, and then air-dried. Finally, 10 μl of 4′,6-diamidino-2-phenylindole (DAPI) antifade solution (1 μg ml‒1) was added to each slide and sealed with a coverslip. Microscopy To detect the ultrastructural chromatin organization of cells at a resolution of ∼120 nm (super-resolution achieved with a 488 nm laser excitation), spatial structured illumination microscopy (3D-SIM) was performed with a 63×/1.4 Oil Plan-Apochromat objective of an Elyra 7 microscope system and the software ZENBlack (Carl Zeiss GmbH, Jena, Germany). Images were captured separately for each fluorochrome using the 405, 488, and 561 nm laser lines for excitation and appropriate emission filters (Weisshart et al., 2016). The spatial animation was done based on the SIM image stack (Supplementary Video S1) using Imaris 9.7 (Bitplane, USA) software. The tool ‘Surface’ was applied to indicate nuclei volumes. Microspores stained with DAPI were counted using the fluorescence microscope AXIO Lab.A1 (Carl Zeiss GmbH). In total, 17 anthers from 17 individual ramets representing the three clones were individually mounted on slides and investigated. Results Microscopic observation and determination of the male meiotic stages Pollen mother cells at the zygotene stage in L. turionifera 9434 and L. minor 5500, and at the pachytene stage in L. turionifera 7432, were observed when anthers just became detectable at sizes of 143×134 μm, 157×119 μm, and 131×91 μm; length×width, respectively. Pollen mother cells of both Lemna species stained with DAPI displayed the typical meiotic stages, such as zygotene, pachytene, metaphase I and II, similar to those of other eukaryotes (Fig. 2). Metaphase I of L. turionifera 9434 and 7432 showed the presence of 21 bivalents, indicating a haploid chromosome number relative to the somatic complement. Meiosis was regular in both species, and the distinct meiotic stages appeared synchronously. Meiotic stages of Lemna turionifera (accessions 9434 and 7432). Zygotene shows that homologous chromosomes start pairing. Chromosome pairing is complete in pachytene. Metaphase I contains 21 bivalents (numbered) in both accessions. Two photos of metaphase II show the chromosome arrangement from different orientations. Scale bars=5μm. Tetrads as the meiotic end-product were detected in developing anthers with mean (±SD) sizes of (175±15)×(144±20) μm for L. turionifera 9434 (n=4), (196±12)×(153±8) μm for L. turionifera 7432 (n=7), and (219±14)×(181±12) μm for L. minor 5500 (n=6). The anther sizes corresponding to the tetrad stage seem variable amongst the three accessions, suggesting that anther size alone cannot serve as an absolute criterion for identifying the developmental stage, even within the same species (L. turionifera). At the tetrad stage of the primary anther, the secondary anther typically exhibited cells at the zygotene to metaphase I stages. Detection of various meiotic products and confirmation of the number of nuclei in each cell At the last stage of meiosis, across all three accessions, most tetrads displayed a tetragonal-like arrangement of the resulting four cells. Flat clover leaf- and T-shaped forms were also observed, albeit at a lower frequency. In addition to tetrads, dyads and triads were also detected. In total, 4710 cells were observed. Lemna turionifera 9434 and 7432 produced 96.9% and 98.8% reduced gametes as well as 3.1% and 1.2% unreduced gametes, respectively. Lemna minor 5500 showed 96.3% reduced and 3.7% unreduced gametes (Table 1). The frequency of different meiotic end-products varied not only between accessions but also among individuals within the same accession. Cells of dyads and their nuclei were noticeably larger, about double the size than those of tetrads (Supplementary Fig. S1 and Videos S1, S2). Thus, tetrad formation is not due to larger meiocytes, compared with those resulting in unreduced dyads, as observed for jason (jas) mutants in Arabidopsis (Yi et al., 2023). Table 1.Numbers and ratios of reduced (1n) and unreduced (2n) gametes derived from diverse meiotic products (dyads, triads, and tetrads) in individual anthers of three Lemna accessions2n gamete number from dyad2n gamete number from triad1n gamete number from triad1n gamete number from tetrad2n gamete rate (%)1n gamete rate (%) Lemna turionifera 943440005446.8593.152004400.4599.5512243483.8396.170004040.00100.00total542417363.1296.88 L. turionifera 74322244200.9399.074005560.7199.290001120.00100.004005800.6899.320004120.00100.004244041.4598.556481884.8595.15total1461215841.2498.76 L. minor 55002003720.5399.474122002.4297.582242681.4598.5532112229212.0487.966003361.7598.250004520.00100.00total40132613483.7196.29 Numbers and ratios of reduced (1n) and unreduced (2n) gametes derived from diverse meiotic products (dyads, triads, and tetrads) in individual anthers of three Lemna accessions FISH revealed the same number of rDNA signals in dyads within their two (or already fused) nuclei, as observed in somatic cell nuclei. However, each tetrad cell nucleus showed only one 45S rDNA signal (Fig. 3). This was to be expected for single loci of 5S and 45S rDNA, as reported for Lemna species so far (Hoang et al., 2019). Therefore, the presence of tetrads, dyads, and a few triads is consistent with the occurrence of reduced and unreduced gametes in the three accessions of the two frequently hybridizing Lemna species. Fluorescence in situ hybridization (FISH) with rDNA detects unreduced gametes in the genus Lemna. (A) A single slice of a 3D-SIM image stack of a somatic interphase nucleus of L. turionifera accession 9434 showing an extended major 45S rDNA signal attached to a nucleolus (red arrow) and a minor signal at the edge of another nucleolus (yellow arrow). Both homologous 5S rDNA signals are marked by green arrows. (B1) Minor 45S rDNA signals (yellow arrows) are visible in the two nuclei of the top cells of a tetrad. Major signals (red arrows) are in the bottom cells. [Copy number heterozygosity between homologous rDNA loci was also documented for Spirodela polyrhiza 5S rDNA by Stepanenko et al. (2026)]. (B2) One of the major 45S rDNA signals in an unreduced dyad is dispersed within the nucleolus of the top cell (cross), the second homologous locus appears condensed outside (triangle). The bottom cell contains the two minor signals of the second homolog (yellow arrows). The small cells adjacent to the dyad are part of the tapetum cells. Both images in (B) are maximum intensity projections (see also Supplementary Videos S1 and S2). Discussion Unreduced gametes, as the most important source of polyploids (followed by early somatic chromosome doubling of diploid zygotes), can arise through multiple cytological pathways (Brownfield and Köhler, 2011). In plants, defective synapsis, abnormal spindle orientation, omission of meiosis I and/or II, impaired cytokinesis, and restitution of nuclei were reported as causes for unreduced gametes (Peloquin et al., 1989; Bretagnolle and Thompson, 1995; De Storme and Geelen, 2011). Consistent with these mechanisms, our observations revealed restitution of nuclei after lacking cytokinesis during male meiosis II in L. minor and L. turionifera. These findings are the first direct evidence showing the proportions of defective products of meiosis for two Lemna species (Table 1, Fig. 3). Studies in Arabidopsis thaliana (L.) Heynh. have identified several core meiotic regulators, such as Arabidopsis thaliana Parallel Spindle 1 (AtPS1), JASON (JAS), Omission of Second Division 1 (OSD1), Cyclin A (CYCA1;2/TAM), SUMO E3 ligase (AtMMS21) and apparently also Structural Maintenance of Chromosome 5/6 (SMC5/6), which, when disrupted, deleted or down-regulated, may cause meiotic defects and result in high frequencies of unreduced gametes (d’Erfurth et al., 2008, 2010; De Storme and Geelen, 2011; Liu et al., 2014; Yang et al., 2021). Based on genomic sequences, Ernst et al. (2025) proposed that L. turionifera and L. minor accessions might exhibit elevated unreduced male gamete production due to deletions in JASON homologs, which are responsible for forming organelle bands and correct cytokinesis during male meiosis II (Brownfield et al., 2015). Our cytological data revealed marked inter-individual variation within this accession (L. turionifera 9434), with unreduced gamete frequencies ranging from 0 to 7% between individuals. Because this is substantially lower than reported for jas mutants of A. thaliana (De Storme and Geelen, 2011), a mutation of JASON as an explanation for the observed variation in Lemna could at best be of low penetrance. OSD and TAM mutations are less likely to be the reason for the observed unreduced meiotic products, because fusion of dyad nuclei indicates that at least division of nuclei during meiosis II was not omitted. Mutations of SMC5/6 components could also mediate formation of unreduced gametes (Yang et al., 2021), albeit it remains unclear how such mutations disturb meiosis II preferentially, because SMC5/6 is mainly required for meiosis I. In wild-type angiosperms lacking obvious mutations in core meiotic regulators, spontaneous unreduced gamete formation is generally rare, typically below 2% in natural populations (Ramsey, 2007; Kreiner et al., 2017). The Lemna individuals examined here showed pronounced variation in unreduced gamete frequencies among individuals of the same clonal accessions, between 0–12% (Table 1). While most individuals produced unreduced gametes at lower frequencies (∼1–5%), a single flower of L. minor yielded 12% (27% of male meiotic end-stages were dyads and triads). Given the rapid clonal propagation characteristic of Lemna, it is possible that somatic mutations affecting meiotic stability were acquired independently among and within different clonal accessions, contributing to this heterogeneity. As an alternative to mutation of meiotic genes, environmental stress causing a switch to sexual propagation, including emergence of a few unreduced gametes, could be responsible for natural auto- and allopolyploids. In vitro flowering requires hormonal treatments that mimic natural stress responses, and such stress has been linked to increased frequencies of unreduced gametes in plants (Ramsey and Schemske, 1998). Under unfavorable conditions that constrain asexual propagation, sexual reproduction could therefore contribute to adaptive genetic diversity, potentially through polyploidization and hybrid formation (Mason and Pires, 2015). Although not easily accessible, unreduced L. minor female gametes should also occur, since triploid L. × japonica with two genomes of L. minor were found, albeit crossing between L. minor × L. turionifera was exclusively unidirectional as was shown by plastid markers (Braglia et al., 2021; Ernst et al., 2025). Despite their predominantly clonal reproduction, L. minor and L. turionifera conform to general patterns reported across angiosperms (Bretagnolle, 2001; Ramsey, 2007), in which a higher frequency of unreduced gametes is restricted to relatively few individuals. Notably, tetraploid L. × japonica has not been reported so far. Low flowering rates of the parental species in natural populations (1.5–5%; Landolt, 1986, pp. 167‒169), together with low frequencies of unreduced gametes, substantially reduce the probability of fusion of unreduced gametes from both parents. This explanation may also account for the absence of autotetraploid L. minor and L. turionifera. Recent evidence from L. × aoukikusa Beppu et Murata suggested that at least some Lemna taxa rapidly absorb the endosperm during embryo development (Lee et al., 2026), indicating that the triploid block may not be of major impact in genus Lemna. It is possible that the dependence on stored reserves provided by the endosperm may be reduced for duckweeds, due to their aquatic lifestyle, which enables photosynthesis immediately after germination of seedlings. Our study provides the basis for comparative cytological analyses of meiotic stages in duckweeds (Figs 1, 2). Diverse natural allopolyploids are even more common in sect. Alatae (Stepanenko et al., 2025) than in sect. Lemna, exemplified by the fertile tetraploid taxon L. × aoukikusa and its triploid backcrosses with L. aequinoctialis Welw. It is most likely that interspecific hybrids also occur in the genus Wolffia Horkel ex Schleid. Thus, knowledge on how to address meiotic processes (this study) as well as species fertility and crossability (Lee et al., 2026) will uncover lineage-specific routes underlying polyploidization and diversification in duckweeds, and provide prerequisites to harness crossing, and thus domestication of economically interesting duckweeds. Supplementary Material