The daylily (Hemerocallis fulva) and nightlily (H. citrina) are typical examples of a butterfly-pollination system and a hawkmoth-pollination system, respectively. H. fulva has diurnal, reddish or orange-colored flowers and is mainly pollinated by diurnal swallowtail butterflies. H. citrina has nocturnal, yellowish flowers with a sweet fragrance and is pollinated by nocturnal hawkmoths. We evaluated the relative roles of flower color and scent on the evolutionary shift from a diurnally flowering ancestor to H. citrina. We conducted a series of experiments that mimic situations in which mutants differing in either flower color, floral scent or both appeared in a diurnally flowering population. An experimental array of 6×6 potted plants, mixed with 24 plants of H. fulva and 12 plants of either F1 or F2 hybrids, were placed in the field, and visitations of swallowtail butterflies and nocturnal hawkmoths were recorded with camcorders. Swallowtail butterflies preferentially visited reddish or orange-colored flowers and hawkmoths preferentially visited yellowish flowers. Neither swallowtail butterflies nor nocturnal hawkmoths showed significant preferences for overall scent emission. Our results suggest that mutations in flower color would be more relevant to the adaptive shift from a diurnally flowering ancestor to H. citrina than that in floral scent.

In flowering plants, pollination success is strongly dependent on the timing of when flowers start to bloom and when they start to close. To elucidate the genetic mechanism influencing the timing of flower opening and closing, we obtained F1 and F2 hybrids of Hemerocallis fulva (a diurnally blooming species, pollinated by swallowtail butterflies) and H. citrina (a nocturnally blooming species, pollinated by nocturnal hawkmoths) and observed their flowering behavior from blooming to closing with the use of digital cameras. For flower opening times, F1 hybrids were highly variable, and F2 hybrids showed a bimodal distribution of flower opening times with peaks in both the morning and evening. The ratio of morning flowering and evening flowering among F2 hybrids did not deviate from 1:1. For the start to close time, both F1 and F2 hybrids were similar in showing the major peak in the evening. The ratio of evening closing and morning closing among F2 hybrids did not deviate from 3:1. These results suggest that the time of flower opening and the start of closing are regulated by different major genes.

Time of flower anthesis in a day is thought to evolve in response to the time of pollinator activities. We studied blooming and withering time in natural populations of daylily (Hemerocallis fulva), nightlily (Hemerocallis citrina) and their hybrids, and also in an artificially obtained array of the F1 hybrids. Blooming time of H. fulva varied from 4:30 to 7:30 and H. citrina varied from 16:30 to 20:30. In a natural hybrid population, blooming time and withering time showed discontinuous bimodal distribution in spite that morphological traits of flowers showed continuous unimodal variation. Most F1 hybrids showed diurnal flowering. These findings indicate that only a few genes have strong phenotypic effect on the determination of flowering time in Hemerocallis, and suggest that the evolution from a H. fulva-like ancestor to H. citrina was not a continuous process by accumulation of minute mutations.

To trace the fate of individual pollen grains through pollination processes, we determined genotypes of single pollen grains deposited on Hemerocallis stigmas in an experimental mixed-species array. Hemerocallis fulva, pollinated by butterflies, has diurnal, reddish and unscented flowers, and H. citrina, pollinated by hawkmoths, has nocturnal, yellowish and sweet scent flowers. We observed pollinator visits to an experimental array of 24 H. fulva and 12 F2 hybrids between the two species (H. fulva and H. citrina) and collected stigmas after every trip bout of swallowtail butterflies or hawkmoths. We then measured selection by swallowtail butterflies or hawkmoths through male and female components of pollination success as determined by single pollen genotyping. As expected, swallowtail butterflies imposed selection on reddish color and weak scent: the number of outcross pollen grains acquired is a quadratic function of flower color with the maximum at reddish color, and the combined pollination success was maximal at weak scent (almost unrecognizable for human). This explains why H. fulva, with reddish flowers and no recognizable scent, is mainly pollinated by swallowtail butterflies. However, we found no evidence of hawkmoths-mediated selection on flower color or scent. Our findings do not support a hypothesis that yellow flower color and strong scent intensity, the distinctive floral characteristics of H. citrina, having evolved in adaptations to hawkmoths. We suggest that the key trait that triggers the evolution of nocturnal flowers is flowering time rather than flower color and scent.

To examine whether floral and post-pollination isolation develops independently or not, we conducted a crossing experiment between Hemerocallis fulva and Hemerocallis citrina that shows large floral divergence adapted for diurnal and nocturnal pollinators that have been believed to be fully cross-fertile. Flowers of the two species from sympatric populations were hand-pollinated with conspecific pollen from the same population (control), interspecific pollen from the same area (sympatric cross), and interspecific pollen from the different area (allopatric cross). After capsule dehiscence, the fruit set, seed set per fruit and seed set per flower were determined among three cross categories. The seed sets per flower were 32 and 77% lower in sympatric and allopatric crosses than in the control when H. fulva was the pollen recipient. There was no difference in three reproductive measures among the cross categories when H. citrina was the pollen recipient. This finding indicates that post-pollination isolation does exist between H. fulva and H. citrina, although it is partial, asymmetric, and weakened in sympatry. Our result suggests that floral and post-pollination isolation may develop independently, and reinforcement may not be a general phenomenon in plants.