All ecotypes of C. crepidioides and C. rubens used in this study, namely, C.c.Ile-Ife, C.c.Osogbo, C.c.Nepal, C.c.Thailand, C.r.Mali, and C.r.Burkina Faso, were described before (Rozhon et al., 2018; Schramm et al., 2021). For cultivation in soil, soil substrate C700 with Cocopor^® (Stender AG, Schermbeck, Germany) was used and the plants were grown in chambers (Bright Boy, CLF Plant Climatics, Wertingen), equipped with Philips Master TL-D 58W/840 light bulbs. The standard growth conditions were a temperature of 25 ± 2°C and cycles of 8 h of darkness and 16 h of white light, with an intensity of 80 μmol m^–2 s^–1. For growth in the greenhouse, the seeds were germinated in the growth chambers, using the conditions described at above, and transferred to the greenhouse for 3–4 weeks, where they were grown at a temperature of 25 ± 3°C in the summer or 20 ± 2°C, and with additional artificial light (80–100 μmol/m^–2 s^–1, for 12 h daily) in the winter, and with a relative humidity of 50%.

For seed analyses, seeds from mother plants grown in the greenhouse were collected at the point of dispersal. They were cleaned, air-dried at room temperature, and stored in paper bags. For germination experiments, seeds from mother plants grown at the same time were sterilized for 20 min in a 75% commercial bleach solution containing 0.01% Triton X-100 and then rinsed 3 times with deionized water. They were plated on half-strength Murashige and Skoog (1/2 MS) medium (Duchefa, Haarlem, Netherland) and incubated in growth chambers at 21 or 25°C in the dark or in long days of 16 h of white light at the required light intensities. The seed germination was assessed after 14 days of incubation and was considered completed when the radical had emerged from the seed coat.

To test the effect of hormones, hormone biosynthesis inhibitors or osmolytes on germination capacities, GA₃ (Duchefa, Haarlem, Netherlands), 24-epiBrassinolide (Apollo Scientific, Cheshire, United Kingdom), (+)-cis,trans-ABA (Duchefa, Haarlem, Netherlands), Fluridone (Sigma, St Louis, MO, United States), brassinazole, and NaCl or D-Mannitol (Sigma–Aldrich, Steinheim, Germany) were dissolved in dimethyl sulfoxide and added at different concentrations to the 1/2 MS medium.

The drought tolerance assays were performed with adult plants with comparable numbers of leaves, using 7-week-old plants of C. rubens Mali and 8-week-old plants of C. crepidioides Ile-Ife, which were grown in the incubators in standard growth conditions (25 ± 2°C and long days of 8-h dark/16 h white light, with an intensity of 80 μmol m^–2 s^–1). The plants were set to comparable watering levels at day 0 and then subjected to progressive drought by withholding water for 7 days before they were re-watered. The survival rates were determined after 3 days of recovery and were defined as the ability to form new leaves.

For detached leaf water loss assays, the third pair of true leaves of 4-week-old plants of C. rubens Mali and 5-week-old plants of C. crepidioides Ile-Ife was detached, placed on an open petri dish, and incubated in room temperature. The leaves were weighed at 1 h intervals for 5 h to determine the rate of water loss.

Crassocephalum crepidioides Ile-Ife and C. rubens Mali were grown for 4 weeks in incubators set to standard growth conditions of 25 ± 2°C and long days (8-h dark/16 h white light, with an intensity of 80 μmol m^–2 s^–1). They were then directly exposed to a temperature of 45°C using an incubator (MIR-154, Panasonic, Japan) and a duration of 2–8 h, before returning to 25°C. Survival was analyzed after 2 weeks of recovery and was defined as the ability to produce new leaves from an intact shoot apical meristem (SAM).

Frost tolerance assays were also performed with 4-week-old plants pre-grown in the Bright Boy incubators at 25 ± 2°C on long days, but by moving them directly to chilling and freezing temperatures of + 4°C to –4°C for 3 h in an incubator (MIR-154, Panasonic, Japan), before returning them to 25°C. Survival was analyzed after 2 weeks of recovery and was defined as the ability to produce new leaves from an intact SAM.

The following chemicals were obtained commercially: ABA and ABA-d₆ (Sigma–Aldrich, Steinheim, Germany); phaseic acid and dihydrophaseic acid (National Research Council Canada, Ottawa, Canada); and ABA glucose ester (Olchemim, Olomouc, Czech Republic). Formic acid was purchased from Merck (Darmstadt, Germany); ethyl acetate and acetonitrile (LC-MS grade) were obtained from Honeywell (Seelze, Germany). Water used for chromatography was prepared using a MiliQ Reference A + Water Purification System (Milipore, Schwalbach, Germany).

Abscisic acid, phaseic acid (PA), dihydrophaseic acid (DPA), and ABA glucose ester (ABA-GE) were quantified from dry seeds, imbibed seeds, and young seedling, and from drought-exposed plants according to Chaudhary et al. (2020) and Abramov et al. (2021). Dry seeds came from seed batches grown in the greenhouse in summer and were grounded directly. To generate material from imbibed seeds, the seeds were imbibed on filter paper for 2 days at 25°C and in the dark. The samples of young seedlings came from 2 to 3-day-old plants with fully expanded cotyledons, which were germinated on filter paper at a temperature of 25°C and grown under long-day conditions (16 h of light/8 h of darkness).

All samples were homogenized in liquid nitrogen and 150–200 mg of material were transferred to a bead beater tube (2 ml, CKMix, Bertin Technologies, Montigny-le-Bretonneux, France), filled with ceramic balls (zirconium oxide; mixed beads with 1.4- and 2.8-mm diameter). Then, ethyl acetate (1 ml) was added, and the samples were spiked with an internal standard solution (20 μl) containing (+)-cis,trans-abscisic acid d₆ (2.5 μg/ml) acetonitrile. Phytohormone extraction was performed by grinding (3 × 30 s with 30 s breaks; 6,000 rpm) using a bead beater homogenizer (Precellys Homogenizer, Bertin Technologies, Montigny-le-Bretonneux, France). The resulting supernatants were membrane filtered, evaporated to dryness, and resumed in acetonitrile (70 μl). Aliquots (2 μl) of all samples were injected into the ultra high performance liquid chromatography (UHPLC)-MS/MS system.

A QTrap 6500 + mass spectrometer (Sciex, Darmstadt, Germany) was used to obtain electrospray ionization (ESI) mass spectra and product ion spectra. The MS/MS system was operated in the scheduled multiple reaction monitoring (MRM) and ESI negative mode. The ion source parameters determined were as follows: ion spray voltage (–4,500 V), curtain gas (35 psi), temperature (550°C), gas 1 (55 psi), gas 2 (65 psi), collision-activated dissociation (-3 V), and entrance potential (–10 V). For the hormone analysis, first, the MS/MS parameters were tuned to achieve fragmentation of the [M-H]^– molecular ions into specific product ions as listed in the following: (+)cis,trans abscisic acid 263→153 (quantifier) and 263→219 (qualifier); (+)-cis,trans-abscisic acid d₆ 269→159 (quantifier) and 269→225 (qualifier); dihydrophaseic acid 281→237 (quantifier) and 281→171 (qualifier); phaseic acid 279→139 (quantifier) and 279→205 (qualifier); ABA glucose ester 425→263 (quantifier) and 425→153 (qualifier).

The mass spectrometer was connected to an ExionLC ultra high performance liquid chromatography (UHPLC) system (Shimadzu, Duisburg, Germany) consisting of two ExionLC AD pumps, an ExionLC degasser, an ExionLC AD autosampler, an ExionLC AC column oven, and an ExionLC controller. Chromatographic separation was achieved with a Kinetex F5 column (100 × 2.1 mm, 100 Å, 1.7 μm, Phenomenex, Aschaffenburg, Germany) operated with a flow rate of 0.4 ml/min and a column oven temperature of 40°C. The solvents consisted of 0.1% formic acid in water (v/v) as solvent A and 0.1% formic acid in acetonitrile (v/v) as solvent B. Chromatography was performed with the following gradient: 0% B held for 2 min, increased in 1 min to 30% B, in 12 min to 30% B, increased in 0.5 min to 100% B, held 2 min isocratically at 100% B, decreased in 0.5 min to 0% B, and held 3 min at 0% B. Instrument control and data acquisition were performed with Analyst 1.6.3 software (Sciex, Darmstadt, Germany). Phytohormone quantification was performed based on comparison with standard curves prepared with reference compounds and expressed as nanogram of hormone per gram of fresh weight of the samples.

For scanning electron microscopic images of buds and flowers, the tissues were shock-frozen in liquid nitrogen, placed on a metal support rack, and visualized with a Hitachi TM3000 scanning electron microscope (Hitachi High-Tech, Tokyo, Japan).

Crassocephalum crepidioides and C. rubens form composite flowers, which are comprised of disk florets that are united in capitula (also called pseudanthia) and are surrounded by involucral bracts (Sakpere et al., 2013). To investigate flower development in some detail, the capitula of two African accessions, C. crepidioides Ilé-Ifè (C.c.Ile-Ife) and C. rubens Mail (C.r.Mali), were analyzed at different stages by light and electron microscopy. This showed that flower development in both species began with the formation of flat buds, which elongated when the florets started to develop inside (stage 1–5; Figure 1A). In bud stage 6, the inner involucral bracts were almost fully developed and the florets started to elongate, with a strong expansion until stage 9 (Figure 1A and Supplementary Figure 1). When this stage was reached, the florets had fully developed and drops of liquid began to be exuded from the tips of the closed capitula (Figure 1B). This liquid was likely nectar secreted from the florets, a phenomenon also found in other plant species with composite flowers (Mesquita-Neto et al., 2020).

Anthesis took place at stage 10. The capitula opened at the top and the purple (C. rubens; Figures 1B–E) or brick–red (C. crepidioides; Figure 1F) tips of the florets showed. Each capitulum contained 120–160 disk florets, which were bisexual and opened from the outside to the inside (Figures 1F–H). The corolla was composed of five petals (Figure 1H) and was tubular, slender, and funnel-shaped. The tube had a white coloration at the base, which transcended to yellow in the upper half (Figure 1D) and ended in a purple (C.r.Mali) or red (C.c.Ile-Ife) limb. Stamens were fused into tubes (Figure 1I,J) with incised bases and purple (C.r.Mali) or red (C.c.Ile-Ife)-colored apexes. The stigma was bilobed (C.r.Mali) or monolobed (C.c.Ile-Ife) (Figures 1D–F) with penicillate tops (Figure 1J).

For fertilization, the mature florets released pollen into the anther column (Figure 1I), which was brushed by the stigma of the growing style to the distal opening (Figure 1J). Once all florets had opened and pollination had occurred, the inner phyllaries defused and the fruits developed.

The fruits were cylindrical dark–brown ribbed achenes with a paler base and apex. They were thinly pubescent and attached to a pappus, which consisted of numerous translucent to white minutely toothed capillary bristles (Figure 2A). The time taken from bud formation to seed release differed between C. rubens and C. crepidioides, with approximately 7 weeks in the case of the former and approximately 6 weeks in the case of the later in greenhouse conditions (Figure 1A).

Interestingly, while similar numbers of florets were formed, much fewer seeds developed in the capitula of C.r.Mali than in those of C.c.Ile-Ife (Figure 2B). To quantify this observation and compare the fertility of C. rubens and C. crepidioides on a broader scale, additional ecotypes were included in the analyses, namely C.c.Osogbo, C.c.Nepal, C.c.Thailand, and C.r.Burkina Faso, and seeds per capitulum were counted. This showed that while C. rubens developed only about 20 seeds/capitulum, almost all florets were fertilized in C. crepidioides, yielding > 100 seeds/capitulum in the analyzed accessions (Figure 2C). The seeds formed by C. crepidioides were smaller and had a lower weight than those of C. rubens (Figures 2B,D).

Crassocephalum crepidioides had previously been shown to have a light requirement for germination (Chen et al., 2009; Nakamura and Hossain, 2009; Sakpere et al., 2013; Yuan and Wen, 2018). To verify this, germination experiments with seeds of C.r.Mali and C.c.Ile-Ife with the same growth history were carried out. Fifty seeds of each accession were sterilized, plated on 1/2 MS medium, incubated in the dark or 16 h of white light/8-h dark at two different temperatures (21 or 25°C), and germination was evaluated at different time points. The results showed that in the dark, seeds of both species exhibited a high level of dormancy. Whereas C. crepidioides seeds had no germination capacity in the dark, C. rubens seeds germinated with approximately 30% efficiency (Figure 2E). Light promoted germination, with stronger effects on C. rubens seeds, which reached 50% germination already after 7 days, whereas C. crepidioides seeds took 10 days at 21°C (Figure 2F). Increasing the temperature to 25°C clearly promoted germination in both species, yielding close to 100% germination after 10 days in the case of C. rubens and after 14 days in the case of C. crepidioides (Figure 2F).

Plant hormones play an important role in seed germination, with gibberellins (GAs) and brassinosteroids (BRs) promoting and ABA impairing it in Arabidopsis thaliana and other plant species (Bentsink and Koornneef, 2008; Unterholzner et al., 2015). To investigate if these hormones may also play a role in the germination of Crassocephalum seeds, the BR 24-epi Brassinolide (epiBL), the GA GA₃, and ABA were added to the 1/2 MS medium in different concentrations and germination outcomes were analyzed. In addition, also inhibitors of hormone activity were applied, namely, Brassinazole (BRZ; Asami et al., 2000) an inhibitor of BR biosynthesis, and Fluridone (Flu; Bartels and Watson, 1978; Kondhare et al., 2014), an inhibitor of ABA biosynthesis. The results showed that whereas epiBL and GA₃ had no significant effect (Supplementary Figures 2A,B), the BR biosynthesis inhibitor BRZ inhibited germination in both species (Supplementary Figure 2C) indicating that BRs promote the germination of Crassocephalum seeds.

Abscisic acid impaired germination in both C.r.Mali and C.c.Ile-Ife, but with much stronger effects on the latter (Figure 3A), for which already a concentration of 0.1 μM ABA reduced germination in the light to 20% at 14 days post-incubation. To analyze this ABA hypersensitivity further, cotyledon greening was assessed, an assay routinely used to quantify ABA responsiveness (Lopez-Molina et al., 2001). This showed that seedlings of C. crepidioides were extremely sensitive to ABA, with 0.1 μM ABA reducing greening of C.c.Ile-Ife to below 20%, an amount were cotyledons of C.r.Mali had a greening rate of 60% after 14 days of incubation (Figure 3A).

These results suggested that in C. crepidioides, ABA biosynthesis and/or responses are increased, which may explain aspects of the low germination capacities of C. crepidioides seeds, especially in the dark. To address, if this may relate to increased ABA biosynthesis, germination outcomes were evaluated in the presence of Flu, which showed that Flu clearly improved germination of both C. rubens and C. crepidioides (Figure 3B). The effects were particularly striking in the case of C. crepidioides, where Flu completely restored germination in the dark, indicating that increased ABA levels may account for the high degree of seed dormancy in this species.

To verify this, levels of ABA were measured by liquid chromatography tandem mass spectrometry (LC-MS/MS) in dry seeds, imbibed seeds, and 3-day-old seedlings, which showed that concentrations of ABA were significantly higher in dry seeds of C. crepidioides than in C. rubens, but dropped to similar levels after imbibition (Figure 3C). Therefore, in summary, seeds of C. crepidioides exhibit a higher level of seed dormancy than those of C. rubens, which is caused by elevated levels of ABA and an ABA hyper-responsiveness and can be broken with the ABA biosynthesis inhibitor Flu, and with light.

In addition to dormancy, ABA plays a major role in abiotic stress responses of plants (Nakashima and Yamaguchi-Shinozaki, 2013) and thus we asked, if the increased ABA responses of C. crepidioides may benefit abiotic stress tolerance of this species. To investigate this, germination and cotyledon greening assays were carried out in the presence of sodium chloride (NaCl) or Mannitol, which are compounds that induce ABA-conferred signaling responses for protection from osmotic stress (Fujita et al., 2011; Yoshida et al., 2014). The results showed that whereas NaCl responses were similar (Supplementary Figure 3), C. crepidioides was hyper-responsive to Mannitol as compared to C. rubens, with particularly clear differences at a concentration of 75 mM (Figure 4), indicating altered tolerance to stress types that induce cellular desiccation.

To investigate this further, detached leaf water loss assays were conducted, to evaluate water loss over time, a parameter that contributes to drought tolerance. The results showed that C. rubens lost water significantly faster than C. crepidioides (Figure 5A), indicating decreased stomatal guard cell conductivity in response to drought. Stomatal conductance is controlled by ABA (Hsu et al., 2021) and to test, if increased ABA concentrations may account for the reduced water loss in C. crepidioides, levels of ABA and its catabolites phaseic acid (PA), dihydrophaseic acid (DPA), and ABA glucose ester (ABA-GE) were measured daily in adult plants of C. rubens and C. crepidioides after water withdrawal for 3 days (Figure 5B). Whereas C. crepidioides exhibited higher ABA levels at day 0, the faster water loss of C. rubens was correlated with a more rapid increase in ABA and its catabolites (Figure 5C), providing evidence that it is not a decreased ability to synthesize ABA, which accounts for the increased water loss of C. rubens.

The ABA measurements showed that the effects of water withdrawal on ABA formation were delayed in C. crepidioides. To assess, if this impacts the drought stress tolerance of the species, a drought stress experiment was carried out, in which water was withheld for 7 days and then the plants were re-watered. Importantly, whereas C. rubens had died off, C. crepidioides recovered (Figure 6), further supporting the idea that it has an elevated drought stress tolerance.

Since ABA can also impact the heat and cold stress resistance of plants (Nakashima and Yamaguchi-Shinozaki, 2013; Huang et al., 2016; Albertos et al., 2022), we compared the ability of C. rubens and C. crepidioides to cope with temperature extremes. While both species survived freezing temperatures of –4°C for 3 h to a similar extent (Supplementary Figure 4), C. crepidioides exhibited a higher heat stress tolerance, surviving 5 h of 45°C, a treatment that C. rubens did not survive (Supplementary Figure 5).