This is the 2D top-down companion to the 3D pollinator-network simulator, computed independently rather than a flattened render of the 3D scene. Plants sit on a top row, pollinators on a bottom row, and a link exists with probability that falls off as a power of each species' rank — exactly how real nested pollination webs are structured (Bascompte et al., PNAS 2003):
p(i,j) = ( 1 - (i/n_p)·(j/n_a) )^γ
n_p, n_a = plant / pollinator counts (ranked generalist → specialist)
γ = nestedness exponent — higher γ concentrates links onto
a shared "generalist core", exactly the nested pattern
seen in real meadows
Connectance C = L / (n_p · n_a)
NODF (approx) = mean pairwise overlap of neighbour sets,
weighted by degree — measures how much a
specialist's partners are a subset of a
generalist's partners
Removing a species can strand its partners with zero remaining links — those go co-extinct too, and the cascade repeats until the network stabilises. A live line chart beneath the network — a genuinely 2D-native view with no equivalent in the 3D version — plots the fraction of species remaining against the attack sequence and shades the area underneath it, so robustness reads as a graph as well as a single number.
R = (area under remaining-fraction curve) / N_total
curve runs from (0, 1) to (removals-to-collapse, 0),
then stays at 0 for the rest of the N_total budget
Note on the robustness formula: this normalises the area by the network's total species count N, not by the number of explicit removal clicks it took to collapse it. Normalising by click-count instead (as a naive port of the 3D version's readout would) makes a one-click catastrophic collapse score the same R≈0.5 as a slow linear decline — the exact opposite of what "robustness" should mean, since a network that falls apart after one removal is not as robust as one that survives many. Normalising by N fixes that: generalists-first collapses fast and scores low, specialists-first survives longest and scores highest, matching the real ecological result.
- Generalists first — removing the best-connected hub species mimics losing honeybees/bumblebees; nested webs are known to be far less robust to this than to random loss.
- Specialists first — removing narrow-niche species barely dents the network, because generalists keep pollinating everyone else.