Zooming in to see the big picture: the value of integrative work to understand early stages of pollinator-mediated plant divergence. A commentary on ‘Geographical variation in flower colour of a food-deceptive orchid reflects local pollinator preferences’

Plant–pollinator interactions literally colour our world through the evolution of floral signals, believed to be key to the diversification of angiosperms. The most fundamental exchange underlying these interactions is a deceptively simple trade, wherein plants offer food and pollinators ensure plants’ reproduction. But what happens when one party reneges on this agreement, as in the case of non-rewarding flowers? One outcome, presented in this issue by Johnson et al. (2025), is floral mimicry, in which non-rewarding, deceptive flowers evolve traits that imitate a model to attract a pollinator (Schiestl and Johnson, 2013). Deceptive pollination, relatively common in the Orchidaceae, provides numerous examples of pollinator-mediated floral evolution, which have fascinated biologists for decades with their ‘various contrivances’ (Darwin, 1877) for luring in pollinators. While the attention paid to these captivating, and often highly specialized, relationships has been foundational to the study of evolution, a disproportionate focus on extreme cases of specialization in plant–pollinator interactions has been argued to obscure the reality of many such interactions in nature (Waser et al., 1996). The Orchidaceae may embody this tension; as one of the most diverse plant families, characterized by a wide array of highly specialized pollination modes, it offers a wealth of opportunities for investigating the role of pollinators in plant diversification. Nonetheless, orchids, particularly deceptive ones, display unique natural history that may not prove widely applicable to other plant systems. For instance, orchids’ large, visible pollinia allow researchers to easily distinguish visitors from effective pollinators and to quantify both pollen removal and deposition (two strengths of the work by Johnson et al. (2025) for understanding fitness implications). Furthermore, this derived floral structure may prevent autonomous self-pollination, thus increasing the likelihood for pollen limitation, associated with strengthened pollinator-mediated selection on reproductive fitness. Secondly, food-deceptive pollination exerts a distinct set of selection pressures on floral signals, as plants must ‘outsmart’ pollinators’ ability to associate their signals with poor rewards, which has been associated with balancing selection that maintains variation in signals to weaken associative learning (Schiestl and Johnson, 2013). While other plants may benefit from pollinator constancy or specialization through reduced stigma clogging, deceptive orchids mimicking food plants encourage pollinators to move indiscriminately between models and mimics, which is likely to alter the evolutionary outcomes of these interactions relative to non-mimetic plants.

In this issue, Johnson et al. (2025) explore how the unique eco-evolutionary paths of mimetic orchids unfold, by investigating intraspecific variation in floral traits of the food-deceptive Disa porrecta in South Africa. Their work reveals shifts in floral colour and morphology driven by a combination of local pollinator preference and attuned mimicry of geographically varying co-flowering plant communities. Geographic variation in D. porrecta comprises two floral forms, either orange-flowered with upward-pointing nectar spurs in most of its range or pink with horizontal orientation at its western range extent. Through field experiments, the authors demonstrate that this geographic variation reflects local adaptation via effective mimicry of different communities of nectar plants and their visitors: orange aphelids in the east, visited by a nymphalid butterfly species, and an assemblage of pink-flowered nectar plants in the west, which comprise a guild pollinated by the long-proboscid fly Prosoeca longipennis. The authors identify convergence in floral colour among these communities, including a gradient from yellow to orange along individual flowers of both models and eastern mimics, implying a fine-tuned molecular basis controlling not only pigment but also patterning and localized expression of colour among ecotypes. Pollinators are shown to prefer local ecotypes in reciprocal translocations: long-proboscid flies display an innate preference for pink flowers within their range, while the nymphalid butterfly, the sole pollinator (and sole reported visitor) of D. porrecta in the eastern range (but present as a co-pollinator in the west), demonstrates preference based on associative learning, preferring Disa flowers that match the colour of locally abundant nectar plants. Thus, this study demonstrates both innate sensory bias (flies for pink flowers) and associative learning (butterflies whose preference shifts based on previous experience of rewarding flowers), two contrasting pathways for pollinator preference (Schiestl and Johnson, 2013), which interact with the ecological context (the co-flowering community and local pollinator assemblage) to facilitate local adaptation of distinct pollination ecotypes in this compelling case.

Additionally, this work’s focus on floral shifts below the species level highlights the value of studying early stages of divergence to understand pollinator-mediated plant speciation. As we continue to amass evidence for the role of pollinators in plant diversification at the macroevolutionary scale, questions remain unanswered about how these influences emerge at recent evolutionary time scales, and how ecological and microevolutionary processes (e.g. intraspecific trait variation, pollinator foraging behaviour, plant mating patterns, and contemporary patterns of selection and gene flow within and among populations) accumulate over time and space to generate macroevolutionary patterns. By leveraging the power of existing intraspecific variation to understand how early stages of divergence first arise, studies below the species level can address persistent questions such as (1) how do pollinator shifts initiate, when shifts in individual floral traits may not occur in concert, potentially resulting in mismatches among floral phenotypes and available pollinators? and (2) are pollinator shifts more likely to occur between pollination modes with similar levels of specialization, or can more specialized systems emerge from within generalized systems (or vice versa)? Also, if so, is ecogeographic isolation required to facilitate such a shift?

To address these questions, unconventional systems may hold promise, such as young lineages with high levels of trait variation and (uncoincidentally) taxonomic uncertainty. Such groups may exhibit ongoing and incomplete divergence, thus affording opportunities to study the formation of incipient barriers in situ. Recent work in one example, the largely North American genus Castilleja, suggests genomic, phenotypic and ecological signals of differentiation are often conflicting (Wenzell et al., 2021), but high degrees of floral trait variation have been linked to a geographic mosaic of pollinator assemblages (Fig. 1A; Wenzell et al., 2023). Several species are visited by largely generalized visitor assemblages despite zygomorphic, restrictive floral tubes. However, the evolution of elongated corollas in one species, C. sessiliflora, appears to filter the pool of potential visitors and increase pollination effectiveness of hawkmoths, but with co-pollination by bees across the range that provide reproductive assurance where hawkmoths are scarce (Wenzell et al., 2024). This and similar examples, along with recent theoretical frameworks, underscores the likely ubiquity of co-pollination systems (Ohashi et al., 2021). Whether co-pollination systems are likely to be stable (when two or more pollinator functional groups serve as effective pollinators without a net loss of fitness – proposed to be the case for pink D. porrecta co-pollinated by flies and butterflies (Johnson et al., 2025)) or when they may be predicted to continue shifting towards specialization (by the evolution of traits to exclude inferior pollinators) or generalization (e.g. due to changes in the distribution or abundance of plants or pollinators) remains uncertain. This challenge emphasizes the continued importance of robust field-based studies (like Johnson et al., 2025) and those spanning the continuum of generalization and specialization in pollination modes.

Fig. 1.

Integrative approaches spanning biological scales and subdisciplines offer complementary – and equally vital – perspectives on early stages of floral divergence. (A) Detailed field studies of geographic variation in floral traits have revealed how pollinator mosaics, including co-pollination systems (top), may drive floral colour variation (bottom), shown here in Castilleja (Wenzell et al., 2023, 2024) and in this issue by Johnson et al. (2025). (B) When studies of intraspecific floral shifts can integrate quantitative phenotyping with controlled pollinator behaviour experiments (top), biochemistry of floral pigments (bottom) and molecular genetics of floral traits, additional insights into the mechanisms of floral adaptation can be revealed, as shown in Mimulus (Wenzell, et al., 2025). Photos: K. Skogen (bottom photo in panel A, modified from Wenzell et al., 2021) and K. Wenzell (modified from Wenzell et al., 2023), shared under a Creative Commons Attribution License (https://creativecommons.org/licenses/by/3.0/; https://creativecommons.org/licenses/by/4.0/).

In addition to looking across systems, future advances require studies that span biological scales to understand processes from molecules to landscapes. The case of pollination ecotypes reported by Johnson et al. (2025) offers exciting opportunities for integrative work into the biochemical and genetic basis of trait variation, further illuminating the mode and mechanisms of pollinator-mediated floral divergence (Frachon et al., 2021). Chief among these opportunities is in floral colour, hypothesized by Johnson et al. (2025) to involve a shift between orange carotenoids and pink anthocyanins, but where additional work could substantially clarify the order and impetus for the shift. Characterizing the individual pigments and genetic mechanisms underlying reflectance is valuable, particularly when mimicry requires convergent evolution of colour and patterning to emulate distantly related model taxa via long-diverged genetic and biochemical pathways. Recent work on intraspecific floral colour shifts, which integrated detailed biochemistry with patterns of expression of structural and regulatory pigment genes, revealed unexpected complexity and convergence in potential incipient pollinator shifts that would not have been uncovered without molecular analyses (Fig. 1B; Wenzell, et al., 2025). Similarly, if genome-wide patterns of variability were explored within and among Disa species, this additional information may clarify the direction of the shift between orange and pink flowers in D. porrecta and related species, compared with the present analysis based on a few highly conserved genomic regions. Currently, Johnson et al. present two compelling scenarios for how the shift from orange to pink – or pink to orange – may have arisen but are unable to reconstruct which is most likely. Additional integrative studies that build on the current wealth of field-collected ecological data with biochemical and genetic analyses could clarify remaining uncertainty in numerous ways. Firstly, by expanding genetic analysis to the genome-wide level, increased genetic variability may inform ancestral state reconstruction, or could reconstruct phylogeographic history to reveal the direction of range expansion in the group. Next, molecular genetic analyses could reveal whether structural and/or regulatory genes underlie colour variation. This work could indicate whether structural pigment genes were gained or lost during floral colour shifts (not a trivial task within a species), or whether they remain present in both colour forms but are differentially regulated at the cis- or trans-regulatory level, allowing rapid (and potentially reversible) shifts. Finally, an enhanced understanding of the biochemical basis of colour shifts (and mimicry) would open the door to further questions, such as: are the pigments underlying pink and orange flowers of D. porrecta conserved within the genus, or do different pigments suggest repeated independent evolution of colour among closely related Disa species? To illuminate how animal pollinators shape the diversity of angiosperms, we must continue to integrate across biological scales, while also valuing the detailed field studies of plant–pollinator evolutionary ecology needed to understand dynamics in situ. In short, future advances will depend on scientific approaches as diverse and varied as the natural world we hope to comprehend.

Funding

K.W. is supported by NSF DBI Award No. 2208984.