At a cutoff of three contacts, axon-to-axon links had 51% higher target overlap than graphs with fixed grouped counts in one fly larva

We tested whether axon-to-axon links join neurons that send output to the same downstream neurons. We used 2,606 labeled neurons from one larval fly brain in Winding et al. (2023). The primary cutoff was three contacts per edge in both axon-to-axon and axon-to-dendrite graphs. Target sets used only the fixed axon-to-dendrite graph. Each control kept every neuron's axon-to-axon partner counts by cell class, side, and direct axon-to-dendrite status in both directions. Mean target overlap was 0.226456 in the recorded graph and 0.149799 in the controls. The ratio was 1.512. The central 95% sampled graph range was 0.148283 to 0.151251. An exact audit found that 7,949 of 8,853 recorded links could change under these constraints. The excess remained at one and five contacts and after removal of listed learning-center classes. Input overlap was also enriched. Thus, the result does not isolate an output-specific rule. The controls omit physical contact and fine cell types. The study uses one animal and does not test synaptic sign, release control, or behavior. The package preserves code, compact edge data, source hashes, all draws, two figures, checks, and repeat steps. A separate run reproduced all 32 main chains exactly.

larval Drosophila · connectome · axon-to-axon · axon-to-dendrite · target overlap · fixed partner counts · alternating cycle · null model