Figure 8.

Non-stochastic Distribution of Putative Synaptic Sites. Monte Carlo statistical analyses reveals that appositions are not the result of stochastic chance overlap related to labeling density. The initial comparison of the subjective identification of contacts along a dendritic segment revealed a low correlation between contact density and the amount of label within a 5μm distance surrounding the dendrite (A) CT: r = 0.19. (B) IX: r = 0.33 indicating that the presence of a subjectively identified putative synaptic site was not simply due to the density of afferent label surround a given dendrite. However, as the presence of a synapse does depend on at least some amount of label being present, a Monte Carlo simulation was performed to determine the stochastic relationship between label density and contact density. (C) A theoretical dendritic cylinder was constructed in Matlab with the three-dimensional voxelated space surrounding it (inclusive of the inner perimeter of the cylinder) corresponding to the dimensions and resolution of the neuropil surrounding our actual dendritic segments (average length and diameter of those segments from A and B). A random assortment of voxels was then selected corresponding to 0.01 to 2% of the total space and the number of voxels falling within the cylinder’s perimeter was counted. The example here contained 0.1% of the voxels randomly selected. (D) The density of voxels overlapping the dendritic cylinder was plotted as a function of the percent of voxels selected and a linear equation fitted along with the 95% confidence interval (shaded region). (E & F) This linear equation (solid grey line) and its 95% confidence interval (dashed black lines and shaded region) were then compared with the observed data from the dendritic segments. Here the density of voxels overlapping the dendrite (rather than subjective contacts) was calculated (see Results for rationale). Both the chorda tympani (E) and glossopharyngeal (F) data displayed significant variance outside of the 95% confidence interval for the simulated model indicating that the voxel overlap was not stochastically determined by the density of label surrounding the dendrite.