A stronger floral scent is not always an advantage. When a flower’s pollen-carrying partner changes, the role of scent may change too. How, then, does the evolution of weaker scent occur?
A study published in 2025 by Kathy Darragh and colleagues examined tropical plants in the genus Costus. In this group, a transition from ancestral bee pollination to reliance on hummingbirds has occurred repeatedly in separate lineages. The researchers investigated whether floral scent changed similarly each time and what was happening at the genetic level.
Collecting airborne floral scent from 30 species
The study collected 100 samples in total from 30 Costus species growing in the field and in greenhouses. Flowers on living plants were enclosed in bags, and scent compounds were collected from flowing air over six hours, from around 9 a.m. to around 3 p.m. Air around the flowers was also sampled to distinguish compounds already present in the background.
The compounds were analyzed using GC–MS, an instrument that separates and examines mixtures of substances. The comparison also incorporated evolutionary relationships among the plants and included eight independent transitions to hummingbird pollination.
For three closely related species, the researchers additionally used RNA to examine how much individual genes were expressed in flowers. Gene presence was examined separately by comparing the genomes of four species. The genes of all 30 species were not studied at the same depth.
Hummingbird-pollinated flowers tended to have less scent
Bee-pollinated flowers had an average of about nine detected scent compounds, compared with about two in hummingbird-pollinated flowers. Chemical diversity and an instrumental measure of emission amount were also higher in bee-pollinated flowers.
The major compounds were terpenes. Rather than being replaced by compounds unique to hummingbird-pollinated flowers, the scent tended to consist of small amounts of a subset of compounds also present in bee-pollinated flowers. In some samples, no flower-derived scent could be detected after background-air compounds were excluded.
However, not all hummingbird-pollinated flowers were scentless, and some bee-pollinated species had little scent. “Not detected” refers to the collection period and analytical conditions used; it does not mean that not a single molecule was emitted.
Differences in scent amount and diversity persisted after accounting for relatedness among species. Analyses reconstructing ancestral states suggested that scent decreased with transitions to hummingbird pollination, while some bee-pollinated lineages developed richer scents than their ancestors.
Differences in floral gene activity rather than gene loss
The researchers focused on genes encoding enzymes that make terpenes. Comparing flowers from three species, they found high expression of these genes in the scent-rich bee-pollinated species. Expression was very low in the hummingbird-pollinated species with little scent.
Had the genes themselves disappeared from the flowers with little scent? Comparing four genomes revealed no tendency for particularly extensive loss of these genes specifically in the two hummingbird-pollinated species.
Together, these findings led the authors to suggest that reduced scent involves lowering how much the genes are used in flowers rather than losing the genes. By contrast, one bee-pollinated species with a particularly rich scent showed gene duplication and high expression associated with scent diversification.
However, the study did not identify the specific DNA differences that alter expression or manipulate genes to demonstrate restoration of scent. Floral RNA was also sampled at only one time point, so activity occurring earlier may have been missed.
There may be a reason not to smell
The authors consider reduced attraction of bees a likely explanation for scent reduction. Once flowers rely on hummingbirds for pollen transfer, continuing to attract their former partners may not necessarily be beneficial.
Even lineages that shifted to hummingbird pollination relatively recently had little scent, and the same pattern occurred repeatedly. The authors suggest that selection favoring reduced scent may have acted, rather than scent simply becoming unnecessary and slowly declining.
However, the researchers did not manipulate scent and directly compare bee visits or pollen-transfer efficiency. It is also not yet sufficiently understood which of the bees and hummingbirds visiting these Costus plants transfers pollen more effectively under which conditions. The explanation that reduced scent discourages bees remains a hypothesis awaiting testing.
The authors also suggest two possibilities for why the genes remain. Gene decay may take time after expression decreases. Alternatively, because terpenes also contribute to plant defense and other functions, the genes may still be needed elsewhere even when their floral role diminishes.
Summary
In Costus flowers, repeated reductions in scent appear to have accompanied evolutionary shifts from bees to hummingbirds as pollinators. Comparisons of the species studied in detail indicate that this change can arise not only through losing scent-producing genes, but also through changing how they are used in flowers.
Even when different lineages evolve similarly low-scent flowers, their genetic paths need not be the same. The authors identify comparison of gene-regulatory mechanisms across independent lineages as a future task. Floral scent has evolved not only toward greater intensity, but also toward weaker signals according to the partners flowers attract.
References
Darragh, K., Kay, K. M., & Ramírez, S. R. (2025). The Convergent Evolution of Hummingbird Pollination Results in Repeated Floral Scent Loss Through Gene Downregulation. Molecular Biology and Evolution, 42(2), msaf027. DOI: 10.1093/molbev/msaf027 (Full original article).
The research was supported by the US National Science Foundation and the Jean H. Langenheim Endowed Chair in Plant Ecology and Evolution. The authors declared no competing interests. Original article: © The Author(s) 2025, published by Oxford University Press, CC BY-NC 4.0.
