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 with respect to 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 ecological and statistical issues directly affect ecologists' abilities to accurately construct ecological networks. However, statistical biases introduced by sampling are difficult to quantify in the absence of full knowledge of the underlying ecological network's structure. To explore properties of large-scale ecological networks, we developed the software 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 mathematical sampling designs that correspond to methods used 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 depends strongly 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, modules with nested structure were the easiest to detect, regardless of the sampling design used. Sampling a network starting with any species that had a high degree (e.g., abundant generalist species) was consistently found to be the most accurate strategy to estimate network structure. Because high-degree species tend to be generalists, abundant in natural communities relative to specialists, and connected to each other, sampling by degree may therefore be common but unintentional in empirical sampling of networks. 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. To reduce biases introduced by sampling methods, we recommend that these findings be incorporated into field design considerations for projects aiming to characterize large species interaction networks.

Fruit traits and temporal abundance shape plant-frugivore interaction networks in a seasonal tropical forest.

Interactions between fleshy fruited plants and frugivores are crucial for the structuring and functioning of biotic communities, particularly in tropical forests where both groups are diverse and play different roles in network organization. However, it remains poorly understood how different groups of frugivore species and fruit traits contribute to network structure. We recorded interactions among 28 plant species and three groups of frugivores (birds, bats, and non-flying mammals) in a seasonal forest in Mexico to determine which species contribute more to network structure and evaluate the importance of each species. We also determined whether fruit abundance, water content, morphology traits, and fruiting phenology are related to network parameters: the number of interactions, species contribution to nestedness, and species strength. We found that plants did not depend on a single group of frugivores, but rather on one species of each group: the bird Pitangus sulphuratus, the bat Sturnira parvidens, and the non-flying mammal Procyon lotor. The abundance, size, and water content of the fruits were significantly related to the contribution to nestedness, number of interactions, and species strength index of plant species. Tree species and birds contributed mainly to the nested structure of the network. We show that the structure of plant-frugivore networks in this seasonal forest is non-random and that fruit traits (i.e., abundance, phenology, size, and water content) are important factors shaping plant-frugivore networks. Identification of the key species and their traits that maintain the complex structure of species interactions is therefore fundamental for the integral conservation of tropical forests.

Influences of sampling effort on detected patterns and structuring processes of a Neotropical plant-hummingbird network.

Virtually all empirical ecological interaction networks to some extent suffer from undersampling. However, how limitations imposed by sampling incompleteness affect our understanding of ecological networks is still poorly explored, which may hinder further advances in the field. Here, we use a plant-hummingbird network with unprecedented sampling effort (2716 h of focal observations) from the Atlantic Rainforest in Brazil, to investigate how sampling effort affects the description of network structure (i.e. widely used network metrics) and the relative importance of distinct processes (i.e. species abundances vs. traits) in determining the frequency of pairwise interactions. By dividing the network into time slices representing a gradient of sampling effort, we show that quantitative metrics, such as interaction evenness, specialization (H2 '), weighted nestedness (wNODF) and modularity (Q; QuanBiMo algorithm) were less biased by sampling incompleteness than binary metrics. Furthermore, the significance of some network metrics changed along the sampling effort gradient. Nevertheless, the higher importance of traits in structuring the network was apparent even with small sampling effort. Our results (i) warn against using very poorly sampled networks as this may bias our understanding of networks, both their patterns and structuring processes, (ii) encourage the use of quantitative metrics little influenced by sampling when performing spatio-temporal comparisons and (iii) indicate that in networks strongly constrained by species traits, such as plant-hummingbird networks, even small sampling is sufficient to detect their relative importance for the frequencies of interactions. Finally, we argue that similar effects of sampling are expected for other highly specialized subnetworks.