COS 210-5 - No strong evidence that modularity, specialization, or nestedness are linked to seasonal climatic variability in empirical bipartite networks on a global scale

Background/Question/Methods

Representation of ecological communities as networks has increased dramatically in the past few decades. While ecologists have long understood that temporal periodicity, and in particular seasonality, are important components of ecological systems, few studies have actually investigated such relationships in empirical communities. Hence, an emerging research area within community ecology aims to understand if and how ecological networks are structured by the seasonal climatic variability they experience. Here, using 723 open-source bipartite networks, we test whether: (1) weighted metrics of nestedness, modularity, and/or specialization vary with temperature or precipitation seasonality in plant-pollinator, seed-dispersal, or host-parasite systems, and (2) temperature and/or precipitation seasonality explains unweighted measures of modularity and nestedness in a redundancy analysis, while controlling for potential differences between antagonistic and mutualistic systems given some evidence that these disparate systems are structured differently.

Results/Conclusions

Altogether, we find that weighted network metrics exhibit only weak relationships with seasonality (i.e., the strongest relationship found had an R²_adj of 0.10), and even while controlling for network type, seasonality does not explain unweighted modularity and nestedness metrics (R²_adj=0.05). Our study also highlights the large amount of structural heterogeneity found in our open-source networks, likely due to the many different sampling and network construction techniques adopted by researchers. Thus, while seasonality may constrain network structure in theory, detecting the nature of such relationships may be extremely difficult to capture in practice. Particularly, practical sampling effects likely inhibit and blur the potential theoretical signal between network structure and seasonality in open-source networks. For example, the duration and amount of geographical used to characterize a community, and the procedure used when collecting ecological data, influences the depiction of a community as a network, all of which varies considerably in networks from open-sources. Hence, a definitive test for the relationship between network structure across large spatial extents likely requires a dataset that is free from these sampling issues, whereby there is a consistent sampling protocol conducted by researchers to characterize communities as networks.