Figure 9. Effect of anisotropy.

(A) Packing volume and network paths. In our simulations increased anisotropy α leads to a reduced number of axons encouters but also a decrease in network efficiency E and an increase in mean shortest path length l. The dashed line marks the mean empirical value of α for collapsed axon distributions and the extent of the shaded region indicates the standard deviation in that quantity. Each data point is the average result for 10 networks, each generated with N  =  2500 nodes, having n_axons  =  10, drawn from under a distribution having anisotropy as indicated and length distributed as a gamma-2 distribution with average length 1000 µm. (B) Node betweenness centrality. Increased anisotropy α leads to an increased right-skewness in the distribution of node betweenness centrality in the generated networks. The dashed line and shaded region, respectively, denote the mean and standard deviation of the empirical values for α. These data were generated using the same network parameters as in (A). See also Figure S3 for a comparison of the histograms of the betweenness centrality distribution at α  =  0.02, α  =  0.7 and α  =  0.9.