Discussion

The Janzen-Connell hypothesis postulates that coexistence in plant communities is maintained by specialized natural enemies that reduce the density of offspring close to conspecific adults (or in high density clumps). Most tests of this hypothesis are indirect without accounting for the plant-associated organisms that drive these patterns (Comita et al., 2014) or are focused on highly host-specific organisms and single plant-pathogen systems following the approach of Augspurger and Kelly (1984). However, little is known about how the entire community of these organisms shapes the spatial structure and assembly of plant communities, and more specifically, how the spatial pattern of these organisms changes during the ontogeny of the plant species and depending on the plants’ seed dispersal mechanism.
Our unique data set on the communities of insect and fungi species inhabiting the leaves of woody plant species in two Mediterranean forests showed that a substantial proportion of organisms is hosted by several plant species. We also found systematic differences between our study forest plots, with organisms at MFJ being more “host-specific” than organisms at MFS. Additionally, our spatial analysis revealed several non-random patterns in the number of plant-associated organisms shared by saplings with their sapling and adult neighbours, and by adults with their adult neighbours. These patterns differed strongly between fleshy-fruited and dry-fruited species and, in most cases, were consistent with expectations for the occurrence or non-occurrence of Janzen-Connell effects in the two seed dispersal mechanism. Our results provide deep insight into how the neighbourhood load of pathogens, epiphytes, and herbivorous insects of the phyllosphere changes during the ontogeny of plant species in Mediterranean forests, and how the resulting spatial arrangements of interactions are built through Janzen-Connell effects and seed dispersal mechanisms. We provide novel evidence about the importance of generalist antagonists and mutualists as drivers of the spatial assembly of plant communities (Sutherland et al., 2013).

Consequences of seedscape and dispersal limitation

Dynamics of fleshy-fruited species at early life-stages

The spatial template created by different seed dispersal mechanisms has consequences for subsequent processes and plant fitness (Howe, 1993; Schupp et al., 2010; Van Leeuwen et al., 2022). Fleshy-fruited species disperse most of their seed by means of frugivorous birds that create a spatial template where seeds are usually located far from the mother plants and in low densities (Howe & Smallwood, 1982; Beckman & Sullivan, 2023). Additionally, frugivores disperse seeds of several fleshy-fruited species together and/or place them often under adults of heterospecific fleshy-fruited species (Jordano 2014; Perea et al., 2021; Verdú & García-Fayos 1996). Such a pattern was found at both communities, where saplings were on average surrounded up to 5m by ~75% of heterospecifics saplings and by ~85% of heterospecifics adults.
Recruitment in neighbourhoods dominated by heterospecifics supresses two parallel mechanisms (density and distance dependency) with antagonists as actors. First, the low density of seeds in highly heterospecific clumps impedes the action of density-dependent mortality agents, allowing the seeding to emerge and get established as sapling (Beckman, 2012; Beckman & Rogers, 2013). We found evidence for this mechanism in our sapling-sapling analysis that showed that saplings in the neighbourhood of saplings did not differ in their loading of antagonists from the null model. Second, frugivore-mediated seed dispersal transports the seeds far from the main antagonist’s sources (i.e. the mother plants), thereby increasing the chance of seed survival and establishment (Howe & Smallwood, 1982). We found evidence for this mechanism in our sapling-adult analysis that showed that adults in the neighbourhood of saplings did not differ in their loading of antagonists from that of randomly placed neighbourhoods.
These two mechanisms are referred to in the literature as distance- and density-dependent processes (Song et al., 2021) and are tested mostly by analysing mortality of early life stages in high density patches or close to conspecifics adults (Janzen-Connell effects) (Comita et al., 2014). In contrast, our evidence is based on the conservation of the spatial template created by frugivorous seed dispersal up to the recruitment stage of the plant species (Perea et al., 2021). This allowed us to confirm the seed escape hypothesis proposed by Howe (1993), since the neighbourhoods of fleshy fruited saplings did not differ from that of randomly selected neighbourhoods (i.e., no significant patterns). However, to fully confirm this hypothesis, we also analysed the dynamics of dry-fruited species that show limited seed dispersal.

Dynamics of dry-fruited species at early life-stages

Seeds of dry-fruited species show usually limited dispersal, leading to strong aggregation of early life stages (Martínez & González-Taboada, 2009) and providing optimal conditions for density-dependent antagonists to proliferate (Howe, 1989). We observed still a high aggregation of saplings (~50% of its neighbours within 5 m were conspecifics; Fig. 3b, d), but this figure reflects most likely density-dependent mortality at earlier stages caused by antagonist because the surviving saplings shared fewer than expected antagonists with their sapling neighbours. Indeed, the proportion of conspecific adult neighbours of saplings was substantially lower than that of conspecific sapling neighbours (between 10 and 30%; Fig. 3f, h), which suggests that spill-over of antagonists from conspecific adults leads to high seed and seedling mortality. In sum, these footprints reflect the density- and distance-dependent Janzen-Connell effects at both studied communities (Bagchi et al., 2014; Prittinen et al., 2003).
Mutualist organisms such as epiphytes (Pajares-Murgó et al., 2022) are expected to enhance species fitness (counteracting the effect of antagonists), but we did not detect footprints of their effects in the sapling-sapling analysis. This suggests that antagonists developed quicker than mutualists in dense seed and seedling clumps (i.e., a priority effect), as reported in other ecological systems (e.g., see Rodriguez-Rodriguez et al., 2015). However, we could identify significant effects of epiphytes in the sapling-adult analysis, but in this case the patterns of both communities were different. Interestingly, saplings shared more than expected epiphytes (i.e. mutualists; Fig. 3h) and more than expected pathogens with their adult neighbours (i.e. antagonist; Fig. 3g). The latter contradicts the hypothesis of the seed escape and our expectations (Beckman & Sullivan 2023), but the pattern of epiphytes was stronger than the pattern of pathogens. This indicates an overall positive balance in the effect of both guilds, as reported in other systems (i.e. soil microbiota - plant-mycorrhiza, Perea et al., 2023). The facilitative effect of epiphytes that weaken Janzen-Connell effects can also explain why saplings at the MFS plot did not show on the first meters a higher than expected proportion of heterospecifics adults (Fig. 4h), as found at the MFJ plot (Fig. 4f) where we did not find facilitative effects of leaf epiphytes.
It remains to explain why facilitative effects emerged at the MFS plot, but not at the MFJ plot. Both forest plots show contrasting species richness and abundance distributions, with MFS owning a lower species richness (17 vs. 29), but a more balanced abundance distribution with four codominant species, whereas dry-fruited Quercus fagineadominates the MFJ plot with 44% of all individuals. Theory suggests that Janzen-Connell effects should be stronger for more abundant species in species-richer communities, a pattern matched at the MFJ plot (Figure S3). Thus, the effects of antagonists may outperform that of mutualists at the MFJ plot, but not at the MFS plot where Janzen-Connell effects are weaker (Liang et al., 2015; Zahra et al., 2021).
Overall, our results support the models proposed by Beckman et al., (2012), and provide new evidence on how seed dispersal and recruit establishment may determine community dynamics (Connell 1971; Janzen 1970). Indeed, seed dispersal matters (Howe & Miriti, 2000), and so does the seedscape, the location where seeds are defecated or regurgitated (Beckman & Rogers 2013; Beckman & Sullivan, 2023; Wenny, 2000;).

Consequences of “seedscape” and recruitment dynamics for community assembly

The spatial patterns found in the adult-adult analysis were consistent between forest plots and depended on the seed dispersal mechanism. Adults of fleshy-fruited species were surrounded by a higher than expected proportion of heterospecifics and by fewer than expected antagonists, whereas no such patterns were found for dry-fruited species. Thus, interestingly, we found reversed patterns compared to the sapling-sapling analysis. While adults of fleshy-fruited species remain at “safe” neighbourhoods characterized by fewer than expected antagonists, adults of dry-fruited species remain at relatively safe neighbourhoods that do not differ in their antagonist load from that of random locations in the plot. Thus, both guilds managed to escape at the adult stage the detrimental effect of a high load of antagonist organisms.