Fig. 4: Moths are innately attracted to humid flowers in a binary-choice behavior assay.

a Setup for the two-choice behavior assay using artificial flowers mounted on two non-flowering Datura plants. Chemical structures indicate the addition of scent (bergamot oil) to both artificial flowers (funnels) at the start of the experiment. Air was pushed through the base of the funnels using an air pump (blue). For humid flowers, the air was pushed via Teflon tubes into a water beaker to generate saturated air, whereas, for ambient flowers, the air was pushed through an empty beaker. Overnight moth visits were video recorded using a motion-sensing IR camera (see Methods). b Measurement of the vertical gradient of floral humidity of the artificial humid flower used in the behavioral experiment. Data were shown for n = 10 transects as mean (solid line) ± SEM (gray shading). c, d Representative images from the videos stored by the motion-sensing camera show moths interacting with flowers. Circles around the flower show the region of interest we drew for position analysis of the moth proboscis (c) and head (d) during probing or entering flower headspace. e–l Behavioral responses of male (slate blue) and female (pale purple) naïve moths for the indicated treatments towards reward-less ambient and humid flowers except for the side-bias test where both flowers presented were ambient humidity. Dots show differences in the duration of proboscis contact and the number of entries in each flower. Red line plots show the mean ± SD. Numbers in parentheses indicate the number of videos in which the labeled body part appeared in the region of interest (c, d) for further analysis. The number of trials for each treatment was as follows: (e, f) Side-bias test n = 12 nights for both sexes; g, h Unmanipulated moths: n = 11 nights for both sexes, i, j Hygrosensor-blocked moths: n = 10 nights for both sexes; k, l Sham control moths: n = 12 nights for males, n = 13 nights for females. For each night of the experiment, 2–4 naïve moths were released in the behavior room. A two-tailed one-sample Wilcoxon signed-rank test against zero was performed on the data. Source data are provided as a Source Data file.