Network structure in wet versus dry seasons within each forest in relation to null models
We predicted that increased seasonality in the rainforest and decreased seasonality in the dry forest, would lead to different impacts on the structure of mutualistic networks. In the dry forest, we found that while the wet season during the El Niño year showed no changes in network structure in relation to the null models, the network in the dry season showed several metrics that deviated from expectations, and that these occurred in different directions. On the other hand, two network metrics during the wet season of the normal year (weighted NODF and weighted connectance) showed similar trends to the dry season and were lower than expected by chance. Explanations for these observations are likely to reflect changes in fruit availability. Notably, in the dry forest, most of the consumption of fleshy fruits occurs at the beginning of the wet season (Vieira & Aldicir, 2006). Although there was a big decrease in the rainfall in the wet season during this study, the second month of the wet season (June) had rainfall within the historical standard deviation that may have restored some of the normal conditions of the forest in terms of fruit availability. This may explain why nestedness (weighted NODF) was low during the dry season and why the network did not show any significant deviation in structure in the wet season. Another possibility is that the drought could have also led to an increase in fruit availability during the beginning of the wet season, which usually happens after a period of drought and leaf fall (Detto, Wright, Calderón, & Muller-Landau, 2018).
During the normal year, the low values of nestedness and connectance are likely explained by the low niche overlap between bat species during this season. The low niche overlap is likely linked with the low number of plant species in the diet of each bat species that was showed by the low value of weighted connectance in relation with the null model. The low niche overlap can promote a decrease in nestedness due to a low overlap in the diet of generalists and specialists. Even though the values of niche overlap were lower than expected by chance during the wet seaon of the normal year, it was higher than the values of the wet season of the El Niño year according to our null model comparison. This suggests that even though our null models did not detect a significant difference during one season only, there was a decrease in relation to what would be expected during a normal year.
In the rainforest, the changes across the seasons were also not consistent, which might also reflect changes in food availability across the year. The lower nestedness and higher modularity of the network during the dry season might be reflecting the bat niche specialization during this season due to a reduction in food availability. On the other hand, the absence of any significance difference in relation to the null models during the wet season might be happening due to an increase in fruit production where bats are exploiting more common resources.