(A) A schematic cross-section (upper panel) and top view (lower panel) of gas sampling. The plate is placed upside-down. In the lower panel, crosses indicate positions of the odor source and dots indicate sampling positions, where numbers represent x-y coordinates. In all positions, the tip of the needle was placed ~2 mm below the surface of the agar; an example of the sampling from (x, y) = (11, 0) is shown in the upper panel. The gas sample was immediately subjected to gas chromatography (GC) analysis. Note that one plate was used for each sampling so that sampling did not disturb the odor gradient. (B) A sample record from the GC analysis of 4 μM 2-nonanone. A single large peak was detected at ~260 s, and the peak height was used for calibration. (C) Calibration curve for 2-nonanone. Each dot represents the average of 3–4 experiments, and data on the log-log plot were fit by two simple regression lines for lower (squares) and higher (triangles) concentrations because of the detector's characteristics (see Materials and Methods). (D) 2-nonanone gradient measured along the x axis (left panel) or at (x, y) = (22, 0) and (22, 15) (right panel) at different periods of the assay. The sampling points are indicated in panel A. The gas (0.2 mL) was sampled at 1, 3, 6, 9, and 12 min. Each data point represents the median ± quartile of 7–9 independent experiments. Notably, the error bar of each measured value was small in the right half of the plate (x ≥ 0), where most of the animals were located during the avoidance behavior. This was likely due to the fact that odor diffusion smooths out positional differences in odor concentration. No significant differences were observed along the y axis at the same time point (right panel; Mann-Whitney test). (E) Another example (the upper animal in the top panel of Figure 1A) of correlations between an animal's behavioral pattern and changes in 2-nonanone concentration.