(A) FlyCircuit PN skeletons co-registered in a standard brain and separated by anatomy group (panels). Each skeleton is plotted in a different color. The number of skeletons is determined by frequency in the FlyCircuit dataset not the number of cells in a single brain. LH, MB lobes and AL shown in darker grey. (B) Example LHN primary neurite cluster, PD2, broken down into its constituent anatomy groups (panels) and cell types (colors). (B’) Using NBLAST to disambiguate cell types and anatomy groups. Dendrogram based on hierarchical clustering of NBLAST scores. Node shape and color indicate different anatomy groups. Leaf color indicates whether each skeleton is from the FlyCircuit dataset or a dye-fill from this study. (C) The number of skeletons, cell types and anatomy groups in the LHN dataset in each primary neurite cluster. (D) Histogram showing the number of skeletons for each LHN anatomy group and cell type. (E) Using NBLAST to match a neuron skeleton to the correct anatomy group. Each skeleton in the dataset was removed, NBLASTed against the rest, mean scores were taken per anatomy group, and anatomy groups ranked. Bar chart shows percentage of skeletons matched to, 1, the correct anatomy group and, >=2, incorrect anatomy groups. (F) Defining a ‘core' set of LHNs. Scatter plot shows cell type plotted against the Log10 of their overlap score (see Materials and methods) with PN termini and the proportion of their dendritic arbor (see Materials and methods) in the standard LH (Ito et al., 2014). Horizontal decision boundary at 50%, vertical decision boundary at 50000, red box, non ‘core' LHNs. The dopaminergic MB input neuron, PPL1-a'3 (Aso et al., 2014a), is flagged in orange as an example of a non-core LHON. Points bounded in blue indicate cell types shown in panel A. Points in chartreuse and dark green (rather than gray) indicate cell types belonging to anatomy groups identified in the Gal4 lines we screened. We made electrophysiological recordings from cell types in dark green.