Figure 4. Plotted as a function of evolutionary divergence (measured by ANI_b) for the star and ladder communities at three depths of WGS coverage (10, 50 and 100×); assembly validation statistics N50 (A) and L50 (B), the degree of genome intermixing S_mixing and its approximate first order derivative ΔS_mixing (dashed lines) (C), lastly graphical complexity H_L (D).

The vertical grey dashed line in each panel marks our operationally defined species boundary (ANI_b = 95%). As evolutionary divergence decreased from easily separable species (ANI_b ≈ 85%) to very closely related strains (ANI_b → 1), assemblies went through a transition from a state of high purity (S_mixing ≈0) to a highly degenerate state (S_mixing ≈1), where many contigs were composed of reads from all community members. A crisis point was observed for small evolutionary divergence (α_BL < 0.2924, ANI_b < 95%), where a sharp change in contiguity (implied by N50 and L50) occured. At very low divergence, N50 and L50 statistics implied that assemblies were recovering, while source degeneracy (S_mixing) monotonically increased. Graphical complexity (H_L) exhibited a similar turning point to L50 and was dominated by graph order |n| (number of contigs/nodes).