Discussion

In our study heterospecific pollen interference did not affect seed set or seed number, i.e. it did not affect whether or not a flower would produce at least one seed or the amount of seeds produced. Rare species had a tendency to produce more seeds, but this trend was not significant. Breeding system did affect seed set, but not in relationship to conspecific or heterospecific pollen treatments, with self-incompatible species less likely to set seed compared to self-compatible species. Lastly, for rare recipients treated with pollen from rare donors, more distantly related recipient-donor species pairs tended to have a lower reduction in seed number compared to closely related recipient-donor species pairs. Hereafter we discuss these results as well as potential ecological and evolutionary implication. In a co-flowering community, we can expect common species to receive pollen from a rare species infrequently, thus an adaptation to that type of pollen is unlikely. On the other hand, a rare species is likely to receive frequently pollen from common species, thus making an adaptation to heterospecific pollen receipt more likely, as predicted by the tolerance-hypothesis (Hao et al. 2023). Indeed, in a study by Arceo-Gomez and colleagues (2016) the authors showed how HP tolerance for a Clarkia species did depend on previous exposure of the population to HP, but rather than acting on the recipient individual, would act on the donor individual, by improving CP performance. On the other hand, such adaptation was not observed for a congeneric Clarkia species, suggesting that adaptation is context- and species-specific. Adaptation could explain the low effect of HP overall in our study species. On the other hand, even though our study species do co-occur and co-flower in nature, the seed-material did not consistently originate from populations co-occurring at local scale, but co-occurring only at regional scale, thus missing potential adaptations at the population level. Further, the (co-occurrence) history of the populations from which the seed material was collected is unknown. Another factor we analyzed is the evolutionary relatedness, measured as phylogenetic distance, between recipient and donor species in interaction with recipient and donor status. While overall no pattern emerged, we could show a decrease of HPI for more distantly related recipient-donor pairs when both were rare. The likely reason for the absence of these patterns for common recipients and rare recipients with common donors could be the lack of close relatives for these groups in our study species set. Indeed, the range for these groups included only phylogenetic distances larger than 189 * 106 years. For closely related recipient-donor pairs, a stronger HPI could be caused by similar recognition systems between recipient stigma and donor pollen grains (Moyle and Nakazato 2010; Moyle, Olson, and Tiffin 2004; Scopece, Widmer, and Cozzolino 2008). For a better understanding of these patterns, a study species set with a broader range should be used. In this study we looked only at pairwise HP interactions, while in a plant community it is likely to have multi-species mixes of HP that are transferred between flowers. Ashman and Arceo-Gomez (2011) performed HP hand-pollinations with mixes up to three species and showed that HPI increased with the number of heterospecific pollen donors. Further, the strength depended on specific species composition Some species are known to produce strongly allelopathic pollen (Kanchan and Chanra 1980), but in our study it did not look like a specific species had a consistent negative effect on all other species in terms of seed number (Figure S 9). For our species, self-incompatible species showed an overall lower seed set, while breeding system did not affect seed number or HPI in any way. Self-compatibility might play a role especially in a natural community, where HP can act through the mentor effect (de Nettancourt 1997), and allowing for self-fertilization even in self-incompatible flowers, with consequent ovule abortion (Lynn, Sullivan, and Galen 2022). In our study we showed how self-incompatible species are less likely to produce seeds even when enough pollen is present. We emasculated our recipient species whenever possible prior to treatment, but due to the small flower size, three out of eight species were left with their anthers to avoid complete flower abortion. These three species (Bupleurum rotundifolium , Fallopia convolvulus and Myosotis arvensis ) are all self-compatible. Thus, one explanation could be that seed set induced by selfing is more secure compared to seed set from outcrossed pollen, despite the genetic advantages of outbreeding (Goodwillie and Weber 2018). Specific flower morphology and in particular a smaller stigma size in restrictive flowers (i.e. flowers with a reduced access to the flower interior) have been shown to reduce HP deposition, while at the same time increasing CP deposition (Montgomery and Rathcke 2012). In our study, we did not analyze the effect of flower morphology or flower traits, since due to our small sample size in species number (eight species in total), species and trait would be confounded. In a natural community, flower morphology would also play an important role in terms of pollinator sharing and flower constancy (the tendency of pollinators to forage on the same flower type, Waser 1986) since some flowers are adapted to specific groups of pollinators and thus sharing among these species is more likely. For example, both Ajuga chamaepitys andFallopia convolvulus , being lip flowers, rely on bumblebees as their most common pollinators (Kuehn et al. 2004). Morphological differences and pollinator type can also drive the direction (for example, from species A to species B but not from species B to species A) and amount of pollen transferred (Fang and Huang 2013). In our study, we used all species both as donors and as recipient, and used a saturated amount of pollen, but this could differ in natural conditions, where pollen limitation is common(Burd 1994; Knight et al. 2005; Larson and Barrett 2000). While interspecific pollen transfer seems to be widespread, the magnitude and the effect on reproductive output are highly context dependent. In line with our findings, several studies showed limited or no effect of HP deposition on seed production (see Morales and Traveset 2008 and references within), while conspecific pollen loss seems to be the more important component of interspecific pollen transfer (Mitchell et al. 2009). Further, the impact of HP deposition has been shown to depend on timing of HP application, with seed set more affected when heterospecific pollen is applied prior to conspecific pollen, compared to a simultaneous mixed application (see Morales and Traveset 2008 and references within) as in this study. While in our study we did not find any strong effect of HP on seed set and seed number, HPI remains an important aspect of co-flowering communities (Ashman and Arceo-Gómez 2013), since it allows species to affect other species without direct competition and at a distance above the direct interactions. In our study we could do a full recipient-donor combination treatment for a total of eight species, which could mean a reduced statistical power when comparing groups within. While using a multi-species approach can be labour-intensive, we suggest to increase the number of species for further studies, or to perform multiple experiments to join the data in a meta-analysis, in order to increase statistical power. In a co-flowering community; we can expect a variable and complex pollination landscape that could promote evolution of flower morphology and avoidance mechanism also due to HPI. At the same time, the lack of stable conditions and the variability of the direction of selection between years and generations (Feinsinger 1987) could hamper adaptation to HP deposition, preventing the emergence of clear patterns. It seems that while mechanisms such as HPI and adaptations to it play a role in shaping plant communities, these patterns are highly variable depending on the context and on the species observed. We conclude that heterospecific pollen interference plays a minor role for rare plant species in our study system. Rather, other factors, like pollen limitation mediated by low pollinator visitation rates, or conspecific pollen loss to heterospecific recipient flowers, are likely to affect rare plant species. Adaptation and species-specific interactions may explain the low overall effect of HPI in our study. The complexity of multi-species interactions and the specific composition of heterospecific pollen mixes may further influence the strength of HPI. Additional research is needed to explore these factors and their implications for both in-sit and ex-situ conservation strategies.