This study describes the specialized brood‐site‐based pollination system of Peltandra virginica Kunth (Araceae) with the chloropid fly Elachiptera formosa Loew and provides experimental evidence that gender‐related changes in floral odor composition synchronize pollinator behavior with the blooming sequence. P. virginica is protogynous and does not self‐pollinate because of a strong temporal separation in sexual function, and it is dependent upon insects for pollination because the spathe completely surrounds the spadix during the pistillate stage. Field observations conducted in central New Jersey and southeastern Pennsylvania showed that E. formosa is closely associated with P. virginica inflorescences. Within the floral chamber, E. formosa adults feed on pollen, mate, and find oviposition sites while the larvae complete their development. Although drosophilid and syrphid flies were collected and reared from inflorescences of both sexual stages, only E. formosa emerged from pistillate‐stage inflorescences and adult E. formosa rapidly transferred fluorescent dye particles between inflorescences of both sexual stages in laboratory enclosures. These findings indicate that this fly species is the primary pollinator at our study locations. Field censuses demonstrated that although E. formosa visited P. virginica inflorescences of both sexual stages with equal frequency, the female flies preferentially oviposited within pistillate‐stage inflorescences. Analysis of floral volatiles with gas chromatography‐mass spectrometry showed that only the spathes emit the floral fragrance and that the composition of the floral volatiles changed during the transition from pistillate to staminate stage. A particularly noticeable change occurred in the emission ratio of the two primary floral odor components, 4,5,7‐trimethyl‐6,8‐dioxabicyclo[3.2.1.]octane and an unidentified chemical analog with molecular weight of 142. The relationship between floral volatile composition and fly oviposition behavior was evaluated using sham inflorescences placed in the P. virginica study population. The sham inflorescences were constructed by covering spadices dissected from mature inflorescence buds with spathes excised from different‐age inflorescences. After a 2‐hr‐long exposure period, sham inflorescences fitted with pistillate‐stage spathes contained more E. formosa eggs than the sham inflorescences with staminate‐stage spathes. However, the highest number of ovipositions occurred in sham inflorescences fitted with spathes excised from mature inflorescence buds. The fragrance emitted by these spathes was composed almost entirely of the two principal odor components. The decline in ovipositions observed in both real and sham inflorescences corresponded to an increase in the emission ratio of 4,5,7‐trimethyl‐6,8‐dioxabicyclo[3.2.1.]octane to the molecular weight 142 analog. This suggests that gravid flies searching for oviposition sites used the ratio of the two primary floral volatile components as an inflorescence gender or age recognition cue.
