Host-specificity of antagonists and mutualists
We find a substantial share of organisms between plant species (Table 1, Figure 2). In general, organisms at MFS (plot with 4 dominant species) were less host specific than those at MFJ (Figure 2), with insects at MFJ showing the highest host-specificity. For example, chewer and sucker insects showed at MFJ a mean overlap of 28 and 29%, respectively, but at the MFS plot mean overlap was 52 and 44%, respectively (Figure 2a, b). In contrast, plant species shared a larger proportion of pathogens and epiphytes, with mean values of, approximately 56% in MFJ, and 67% in MFS. Note that we used different methods for species identification within guilds (i.e., insects vs. fungi), which does not allow for a robust comparison among these guilds, but for robust comparison between forest plots.
We found that the mean number of shared organisms within distancer tended to decline with the corresponding proportion of heterospecifics if the organisms showed relatively strong host specificity (Figure S1). This relationship was especially marked for insects at the MFJ plot, which showed the highest level of host specificity. Interestingly, epiphytes at the MFJ plot and epiphytes and pathogens at the MFS plot were less host specific, and showed different patterns (Figure S1).

Shared antagonists and mutualists between neighbouring saplings

As expected, the sapling “seedscape” varied between fruit type, but were similar between plots (see table 1 for expectations). Saplings of dry-fruited species were generally surrounded by other saplings that shared fewer than expected organisms of all four types, with weaker effects for pathogens and epiphytes at the MFJ plot (Figure 3 a-d, grey lines). The seedscape of saplings of fleshy-fruited species, in contrast, did not differ significantly from the expectations of the null model (Figure 3a-d, red lines) (see Table S3 for significant departures from the null model, and figure S2 for graphs without the standardisation).
The contrasting patterns between fleshy- and dry-fruited species are most likely related to two effects. First, saplings of dry fruited species showed a significantly higher than expected proportion of heterospecific saplings in their neighbourhood (Figure 4b, d), and second, they showed very high local dominance of approximately 50%. This result is consistent with a density-dependent seedling mortality caused by Janzen-Connell effects. In contrast, saplings of fleshy-fruited species had approximately 83% (MFJ) and 75% (MFS) heterospecific neighbours (Figure 4a, c). The high levels of heterospecific neighbours are consistent with frugivore seed dispersal of early density-dependent mortality that prevented emergence of Janzen-Connell effects (see Table 1).

Shared antagonists and mutualists between saplings and neighbouring adults

The patterns of similarity in shared organisms between saplings and neighboured adults strongly differed from that revealed in the sapling-sapling analysis (Figure 3). Most notably, the strong negative sapling-sapling effects of dry-fruited species almost completely disappeared, in accordance with our expectations (see table 1); except for shared insects at the MFJ plot up to the 1m neighbourhood (Figures 3e, f). This pattern was caused by a significantly lower than expected proportion of conspecific adults around saplings (Figure 4f), a pattern which was not found in the other 3 cases (Figures 4e, g, h), but probably associated with distance-dependent Janzen-Connell effects.
Interestingly, saplings of dry fruited species at the MFS plot were surrounded by adults sharing significantly more pathogens and epiphytes than expected (Figures 3g, h), whereas they were surrounded by saplings sharing significantly less pathogens and epiphytes than expected (Figure 3c, d). This pattern was probably caused by (slightly) fewer than expected heterospecific neighbours (Figure 4h, Table S3). Thus, the MFS adult “seedscape” surrounding saplings showed large differences depending on whether locations close to other saplings or close to adults are considered. In contrast, at the MFJ plot (the plot with one dominant species), patterns in pathogens and epiphytes did much less differ for dry-fruited species between sapling-sapling and the sapling-adult analysis (i.e., weak or no significant effects). This suggest that different mechanisms operate at the two plant communities, which may be a consequence of the differences in species diversity and dominance (See Figure S3). Similarly, the patterns of fungal guilds for fleshy-fruited species were for both, sapling-sapling and sapling-adult analyses basically non-significant (see Table S3 for significant departures from the null model).

Shared antagonists and mutualists between neighbouring adults

The results of the adult-adult analyses differed substantially from that of the sapling-sapling and sapling-adult analyses. In most analyses we found that adults of fleshy fruited species shared, consistently between plots, fewer than expected antagonists with neighbour adults (red curves in Figure 3i-l); and showed a higher than expected proportion of heterospecific adults in their neighbourhood (Figure 4i and 4k). However, no effects were found for dry-fruited species (grey lines in figure 3i-l). Interestingly, this pattern is opposite to that of found in the sapling-sapling analysis, and points to the importance of seed dispersal as modulators of Janzen-Connell effects.