The effect of network size and sampling completeness in depauperate networks

Area of Science:

• Ecology
• Network Analysis
• Ecological Network Structure

Background:

• Accurate estimation of interaction network structure is crucial for understanding ecosystem stability and function.
• Under-sampling is a common challenge, and its effect on network metrics is increasingly studied.
• The influence of network size on the relationship between sampling completeness and structural metrics remains unclear, impacting comparability.

Purpose of the Study:

• To investigate the combined effects of network size and sampling completeness on ecological network structure.
• To understand within-system variability in metric bias due to under-sampling.
• To assess the comparability of network metrics across systems of varying sizes and sampling levels.

Main Methods:

• Utilized a host-parasitoid model system with spatially distinct subwebs.
• Employed richness estimates to quantify sampling completeness for species and interactions.
• Tested the impact of network size and sampling completeness on twelve unweighted and weighted network metrics.

Main Results:

• Network metric robustness to under-sampling strongly correlated with network size.
• Sampling completeness of interactions was a more reliable predictor of metric bias than species sampling completeness.
• Weighted metrics outperformed unweighted metrics at low sampling completeness, particularly in larger networks.

Conclusions:

• Under-sampling significantly affects the comparability of both unweighted and weighted network metrics, especially in small and size-variable networks.
• Under-sampling poses challenges for within-system comparisons in species-poor networks and can mask issues in species-rich networks.
• Addressing system-specific variations in metric bias is essential to mitigate the impacts of under-sampling in ecological network studies.