Scaling of nestedness in complex networks

Nestedness characterizes the linkage pattern of networked systems, indicating the likelihood that a node is linked to the nodes linked to the nodes with larger degrees than it. Networks of mutualistic relationship between distinct groups of species in ecological communities exhibit such nestedness, which is known to support the network robustness. Despite such importance, quantitative characteristics of nestedness is little understood. Here we take graph-theoretic approach to derive the scaling properties of nestedness in various model networks. Our results show how the heterogeneous connectivity patterns enhance nestedness. Also we find that the nestedness of bipartite networks depend sensitively on the fraction of different types of nodes, causing nestedness to scale differently for nodes of different types.

💡 Research Summary

The paper investigates the scaling behavior of nestedness—a structural property that quantifies how likely a node is to be linked to the neighbors of higher‑degree nodes—in several canonical complex‑network models. Nestedness is measured using the mean topological overlap, defined as the average over all node pairs of the number of shared neighbors divided by the minimum of the two degrees. This metric is computationally simple and captures the intuitive notion of “specialist” nodes linking to the interaction partners of “generalist” nodes.

The authors first assume that the probability of a link between nodes i and j can be factorized as f_{ij}=2L P_i P_j, where L is the expected total number of links and P_i is the selection probability of node i, ordered by decreasing expected degree. Under this assumption the ensemble‑averaged nestedness becomes S = (2L)/(N(N‑1)) I₁ I₂, with I₁ = Σ_{i>j} P_j and I₂ = Σ_ℓ P_ℓ². Consequently, the decay of S with system size N is governed entirely by how fast P_i declines with i.

Three network families are examined:

1. Static scale‑free model – Nodes are selected with probability P_i ∝ i^{‑α} (0 ≤ α < 1), yielding a degree distribution exponent γ = 1 + 1/α. For α < ½ (γ > 3) the factorized form holds for all pairs, giving I₁ ∼ N^{2‑α}, I₂ ∼ N^{‑1} and thus S ∼ N^{‑1}, the same scaling as Erdős–Rényi random graphs. For ½ ≤ α < 1 (2 < γ ≤ 3) the connection probability saturates for high‑degree nodes, leading to I₂ ∼ N^{‑2α} (or N^{‑2α} ln N at the marginal α = ½). The resulting nestedness scales as S ∼ N^{‑2(1‑α)} or S ∼ (ln N)/N, i.e., it decays much more slowly than N^{‑1}. The analytical predictions are confirmed by extensive simulations, showing that heterogeneous degree distributions strongly enhance nestedness for finite networks.

2. Barabási–Albert (BA) growth model – New nodes attach preferentially to existing nodes with probability proportional to degree. The expected degree of a node introduced at time i evolves as k_i(t) = m (t/i)^{½}, leading to a factorized connection probability with P_i ∝ i^{‑½}. This case corresponds to the static model with α = ½, and the nestedness follows S ≈ (2 m³ ln N)/N. Simulations agree with the logarithmic correction, although the prefactor deviates slightly due to dynamic correlations between adjacency entries that are ignored in the factorized approximation.

3. BA‑type bipartite networks – Two node classes (e.g., animals A and plants P) grow simultaneously, each with its own attachment exponent (α_A, α_P) and relative abundance x_A, x_P (x_A + x_P = 1). The nestedness expression splits into contributions from each class, S ∝ (2L)/(N(N‑1))