Jacobaea paludosa is a perennial rhizomatous helophyte, 50–250 cm tall, with an erect fistulous stem and linear-lanceolate leaves with narrowly toothed margins. The inflorescence is a terminal corymb of capitula with peripheral ligulate bright yellow flowers (only female) and tubulous yellow-brownish flowers (hermaphroditic); the fruit is a cypsela provided with a pappus of whitish hairs (Zàngheri 1976; Michna 2006; Pignatti et al. 2017–2019). Italian populations are currently attributed to the subsp. paludosa (Portal to the Flora of Italy 2025) following Calvo et al. (2015), despite marked morphological variability being frequently observed.

J. paludosa has a Eurosiberian distribution and lives in marshes, lakes, riparian forests or in sparse Phragmites australis (Cav.) Trin. ex Steud. reed thickets, as well as on river or canal banks and in wet grasslands with peaty soil (Michna 2006; Persico 2009; Pignatti et al. 2017–2019). In Italy, it grows in various northern regions, but only in sparse, fragmented and often very small unstable populations (Buldrini 2012; Pignatti et al. 2017–2019).

J. paludosa propagates mainly vegetatively through rhizome fragmentation, while seed recruitment appears to be limited in the wild (Michna 2006). Seed viability and germination are highly variable among European populations and tend to be low in small, isolated populations (Walters 1974; Wells et al. 1993; Winter et al. 2008). Similarly to other Asteraceae, J. paludosa seems to be able to self-pollinate to some degree, however results from previous investigations are contradictory (Wells et al. 1994; Michna 2006); a pollination mediated mostly by insects was suggested by Klotz et al. (2002). In Italy, some preliminary observations showed good pollen vitality (50% minimum) and variable seed fertility (9.3 to ca. 50%, unpublished data), with germination ability dropping to zero after about one year.

Breeding system and selfing ability were investigated in a natural population located in Piuda, within the protected area SCI-SAC IT20B0017 ‒ Ansa e Valli del Mincio (45°10'16"N, 10°42'9"E, alt. 19 m a.s.l.; hereafter NAT – Fig. 1). There, the population of J. paludosa counts many hundreds of flowering scapes and grows in a humid grassland with peaty soil, extending over approximately 6.8 ha, characterised by Molinia caerulea (L.) Moench, Carex spp. and various threatened species.

We also selected a second population, along the right bank of the «Collettore delle Acque Basse Reggiane» canal, within the protected area SCI-SAC IT4030015 ‒ Valli di Novellara (44°53'15"N, 10°45'32"E, alt. 18 m a.s.l.; hereafter REI – Fig. 1). Here, individuals of J. paludosa form two distinct nuclei, less than 100 m apart, of about 10–15 scapes each; one nucleus is natural, the other derives from a reinforcement intervention performed in 2006, with seeds taken from the first nucleus (Dallai et al. 2010).

Both populations are subject to seasonal water table level fluctuations, with nearly constant soil imbibition and occasional submersion of the basal part of the plants.

To investigate the species’ self-compatibility and ability to self-pollinate without insect pollination and to assess reproductive success in natural conditions, we performed a pollination experiment in the field, both in NAT and in REI, comparing three tests: a) spontaneous self-pollination, where no manipulation was performed on capitula previously bagged with non-woven fabric bags at bud stage (SS, performed on 6 distinct capitula in NAT and 3 in REI); b) hand self-pollination, where capitula were pollinated manually with pollen from the same individual plant and re-bagged after pollination (HS, performed on 11 capitula in NAT); c) natural pollination, where no manipulation was performed on marked unbagged capitula, freely accessible to floral visitors (NP, in 2016 performed on 6 distinct capitula in NAT and 4 capitula in REI, and in 2017 on 12 distinct capitula in NAT). After withering, all treated capitula were re-bagged to prevent fruit loss or predation. Both NP and SS were performed in NAT and REI populations, whereas HS was implemented only in NAT population (Table 1). We point out that separate scapes of J. paludosa may belong to the same genet rather than being distinct individuals, due to predominant vegetative propagation.

After performing tests, all capitula subject to manipulation were monitored until withering, after which cypselae were collected when ripe.

The embryo vitality test (TTC) was carried out on seeds obtained through SS and NP treatments to assess viability. In particular, we tested whether the morphological appearance of the cypselae was a reliable criterion for predicting viability. All cypselae produced within a capitulum were first attributed to distinct morphological classes: well developed and well-shaped, pale brown in colour (G1 – viable); small, shrunken, and dark coloured (G3); intermediate forms (G2). According to preliminary observations, we expected that: G1-type cypselae would bear a vital embryo, G2-type would be empty or with an unviable embryo, G3-type would always be sterile (without embryo). Following the experiment, morphological categories «viable» and «unviable» were later assigned to the classes.

The TTC test was performed on a subset of seeds from each morphological class (20% of the total number of cypselae per capitulum, following Bacchetta et al. 2006), using 2,3,5-triphenyltetrazolium chloride diluted to 1% in a 0.1 M solution of phosphate buffer (Van Waes and Debergh 1986). Before being immerged in TTC solution, the cypselae were scarified mechanically. Once divided according to morphological class, seeds were placed into a test tube with 6 ml of reagent, after which the tubes were put in a water bath at a constant temperature of 30 °C for 48 hours. Next, the tubes were extracted from the water bath, the TTC solution substituted with distilled water after which the tubes were placed into a fridge (at 4 °C). Embryo vitality was inspected under an optical microscope at a later stage.

To gather information on reproductive success and pollination efficiency, we performed a pollen supplementation experiment in NAT. We marked 12 capitula chosen randomly, and left them unmanipulated (test NP – natural pollination); in parallel, we selected 11 capitula and supplemented flowers with pollen collected from different individuals within the same population (test Xe-intra – intra-population xenogamous pollination). Pollen was collected from different plants, mainly from capitula with abundant exposed pollen. Pollen supplementation was repeated for three consecutive days, in order to pollinate all flowers of the inflorescence. A minimum distance of 5 meters between pollen donors and pollen receiving scapes was kept while performing the test Xe-intra, to limit the risk of intra-genet pollination. Using a clean paintbrush, pollen was applied to flowers exhibiting exposed stigmas. Experimental capitula were monitored until withering and fruiting and cypselae collected once ripened.

To check for possible intra-population inbreeding depression, we performed an inter-population pollination test (Xe-inter) on 11 capitula. The scapes were marked and bagged at bud stage with non-woven fabric as protection against insects. Manipulations were performed as described for Xe-intra, but pollen donors were selected among individuals belonging to REI. To preserve pollen viability, donor capitula were kept in a Falcon tube and maintained in a cool bag during the transfer. Before performing hand pollination, bagged capitula were delicately shaken, to avoid (at least partially) issues deriving from stigma clogging, sensuKawagoe and Suzuki (2005). As for the other experiments, capitula were monitored until withering and fruiting and mature cypselae were collected.

Subsequently, germination ability of viable cypselae (G1) produced under the different treatments was evaluated in the laboratory. Since no specific germination protocol of J. paludosa is available, we adopted the procedure currently in use in Bologna University’s laboratory of agriculture for Jacobaea maritima (L.) Pelser & Meijden (http://www.scienzeagrarie.unibo.it). The test was carried out on fruits collected in 2017, set after the following tests: NP, HS, SS, Xe-intra, Xe-inter. A total of 15 visually well-developed fruits per capitulum were randomly chosen among those collected in the field. All G1 cypselae produced by one capitulum were selected if less than 15 G1 cypselae were available per capitulum. Sample sizes are shown in Table 1. Germination tests lasted a total of 21 days and were considered done when, after two consecutive inspections, no new radicles were visible.

To minimise mould formation, all materials were sterilised using either an autoclave or an alcohol bath (e.g. plastic Petri dishes). Cypselae were surface-sterilised by immersion in 1% NaClO for 3 minutes, rinsed with deionised water for 10 minutes (Rossi et al. 2016), and stored at 4 °C for 5 days, to induce dormancy-breaking thermal shock. Afterwards, cypselae were placed on filter paper moistened with deionised water in Petri dishes and kept in phytotron (12 h light/dark, 20 °C). Filter paper moisture was checked daily, and germination was monitored every 2–3-days. Seeds were considered germinated when radicles were visible to the naked eye.

We performed χ² tests by means of chisq.test function, with p-values adjusted using the Bonferroni correction. In particular, the first analysis involved testing to verify any difference in embryo vitality between G1 cypselae and the sum of G2 and G3 cypselae (ratio G1/(G2+G3)), basically viable/unviable cypselae. Next, we tested the ratio set after NP, SS, HS, Xe-intra and Xe-inter tests, considering pairs of ecologically relevant comparisons. Finally, we compared G1 cypselae germination rates with distinct pairs of tests, considering the numbers of germinated versus the number of ungerminated seeds (sensu lato, including G3 cypselae). All statistical elaborations were performed in RStudio version 4.0.2.

Data available from the Zenodo Digital Repository: https://doi.org/10.5281/zenodo.15461682.

The mean number of cypselae produced per capitulum in NAT is reported in Table 2.

The highest number of G1 cypselae was obtained after NP treatment. HS treatment mainly produced G2 and G3 cypselae, whereas SS treament produced only G2 and G3 cypselae.

The TTC test confirmed that only G1 cypselae carried a vital embryo, with highly significant differences in embryo vitality between G1 and G2+G3 (χ² = 37.38, ν = 1, P < 0.001).

The proportion of viable cypselae produced differed among treatments, resulting significantly higher in NP compared to both SS (which set exclusively G2 and G3 types; χ² = 570.34, ν = 1, P < 0.001) and HS (χ² = 552.44, ν = 1, P < 0.001). However, HS produced significantly more G1 cypselae compared to SS (χ² = 64.78, ν = 1, P < 0.001), which produced none.

Germination rates of G1 cypselae obtained through NP and HS were not significantly different (χ² = 0.325, ν = 1, P = 0.569).

The number of viable fruits produced in natural conditions (NP) was significantly different in NAT and REI populations (χ² = 15.51, ν = 1, P < 0.001), with more G1 cypselae obtained in the former (74.56%) compared to the latter (65.81%). However, the germination of NP-seeds was significantly higher for REI, than for NAT (χ² = 166.68, P < 0.001).

The number of G1 cypselae per capitulum did not significantly differ (χ² = 0.725, ν = 1, P = 0.395) between NP (60.56%) and Xe-intra treatments (72.44%). Equally, no statistical difference was found for germination rates of NP and Xe-intra seeds (χ² = 0.170, P = 0.680). The number of G1 cypselae obtained through Xe-intra was significantly higher than the one obtained through Xe-inter (χ² = 65.67, ν = 1, P < 0.001). However, seeds produced through Xe-inter showed significantly greater germination rates than those produced through Xe-intra (χ² = 4.821, P = 0.028). All ecologically relevant pairwise comparisons and the results of the χ² goodness of fit tests are summarised in Table 3.

The evidence gathered in this study from two peripheral populations of J. paludosa provides useful indications for conservation actions in lowland ecosystems.

Italian populations of J. paludosa are fragmented and isolated, likely characterised by a low genetic diversity, as it often happens in populations located at the range limits (e.g. Paul et al. 2016). Such populations, other than being isolated from one another, are separated from central European ones by the Alps (Conti et al. 2005). Studies carried out on Jacobaea vulgaris Gaertn. reported that seed dispersal occurs within 15 m from the mother plant (McEvoy and Cox 1987), thus making spontaneous reinforcement of Po Plain populations from beyond the Alps very unlikely.

Altogether, our results indicate that the reinforcement intervention performed in 2006 was successful. Regarding the study population, however, a similar intervention may not be sufficient. In fact, seed germination ability in this population is extremely low, indicating that a similar approach would likely be effective using material sourced from a distinct population. In addition, this intervention alone may not be enough to prevent population loss if nothing is done to maintain the original habitat of hydro-hygrophilous species. In fact, if J. paludosa typically grows on unstable floodplains, subject to alternating phases of erosion and deposition, which create and destroy new habitats continuously, what remains of the species in wetlands and marshes would be guaranteed by maintaining traditional management involving harvesting, grazing, fires and alteration of grass cover, imitating natural processes (Michna 2006). This kind of management is becoming rarer nowadays. Furthermore, the decline of J. paludosa is due not only to the ever-increasing deterioration of wetland areas, but also to the lack of regeneration niches in its natural habitat (Diekmann and Bartels 2012). Therefore, alongside with targeted conservation measures, the implementation of policies aimed at preserving and restoring wetland habitats are of fundamental importance.