Revealing biases in the sampling of ecological interaction networks
The structure of ecological interactions is commonly understood through analyses of interaction networks. However, these analyses may be sensitive to sampling biases in both the interactors (the nodes of the network) and interactions (the links between nodes), because the detectability of species and their interactions is highly heterogeneous. These issues may affect the accuracy of empirically constructed ecological networks. Yet statistical biases introduced by sampling error are difficult to quantify in the absence of full knowledge of the underlying ecological networks structure. To explore properties of large-scale modular networks, we developed EcoNetGen, which constructs and samples networks with predetermined topologies. These networks may represent a wide variety of communities that vary in size and types of ecological interactions. We sampled these networks with different sampling designs that may be employed in field observations. The observed networks generated by each sampling process were then analyzed with respect to the number of components, size of components and other network metrics. We show that the sampling effort needed to estimate underlying network properties accurately depends both on the sampling design and on the underlying network topology. In particular, networks with random or scale-free modules require more complete sampling to reveal their structure, compared to networks whose modules are nested or bipartite. Overall, the modules with nested structure were the easiest to detect, regardless of sampling design. Sampling according to species degree (number of interactions) was consistently found to be the most accurate strategy to estimate network structure. Conversely, sampling according to module (representing different interaction types or taxa) results in a rather complete view of certain modules, but fails to provide a complete picture of the underlying network. We recommend that these findings be incorporated into field sampling design of projects aiming to characterize large species interactions networks to reduce sampling biases.\n\nAuthor SummaryEcological interactions are commonly modeled as interaction networks. Analyses of such networks may be sensitive to sampling biases and detection issues in both the interactors and interactions (nodes and links). Yet, statistical biases introduced by sampling error are difficult to quantify in the absence of full knowledge of the underlying networks structure. For insight into ecological networks, we developed software EcoNetGen (available in R and Python). These allow the generation and sampling of several types of large-scale modular networks with predetermined topologies, representing a wide variety of communities and types of ecological interactions. Networks can be sampled according to designs employed in field observations. We demonstrate, through first uses of this software, that underlying network topology interacts strongly with empirical sampling design, and that constructing empirical networks by starting with highly connected species may be the give the best representation of the underlying network.