The study of factors and processes that involve evolutionary divergence can inform how lineages split, which is in turn an important player in speciation, a fundamental process in the generation of diversity (Coyne and Orr 2004; Butlin et al. 2010; Kulmuni et al. 2020). Biological systems undergoing active divergence or incipient speciation are therefore of great importance, as they offer unique opportunities to the study of such processes and factors (Kulmuni et al. 2020).

In ring-species, evolutionary divergence occurs from an ancestral population along two fronts that conform to a circular distribution such that when the terminal populations at either end of the diverging fronts come together, barriers preventing homogenizing gene-flow are already in place so that the two fronts remain as distinct (Irwin et al. 2001). The plant Euphorbia tithymaloides is an example of such system, having colonized the Caribbean Basin from Mexico/Guatemala along two expansion routes: one that travelled east through the Greater Antilles (GA front), and another that moved south through Central America, then east through the northern portion of South America, and then north, traversing the Lesser Antilles (LA front). Models of genetic variation and phylogenetic inference support evolutionary divergence in this system (Cacho and Baum 2012; Cacho et al. 2019), so that in the vicinity of the US Virgin Islands, close to the Anegada Passage, the terminal populations from either front (GA, LA) are the most distantly related and genetically distinct despite occurring in neighbouring geographic areas. Morphologically, the two terminal forms exhibit contrasting leaf morphologies (Jacobo-Arteaga et al. 2022) and habit, as well as some key differences in floral morphology (mainly size and pubescence) but also important similarities (i.e. floral shape; Fig. 1). Vegetatively, individuals from the GA front are small shrubs 0.8–3 m tall that divide at the base giving rise to somewhat delicate scandent branches (1–3 cm thick), and thin, linear and pubescent leaves (Dressler 1957). Individuals of the LA front are robust, self-supporting shrubs or even small treelets (also 1–3 m tall) with a single main stem that can be as wide as 6–9 cm at the base, and thick and glabrous leaves (Dressler 1957) (Fig. 1). Plants from the GA and LA fronts also differ in floral morphology, with those form the GA side presenting smaller and somewhat pubescent cyathia (modified inflorescences of Euphorbia) that produce slightly less pollen (Cacho and José-Zacatula 2020) than those of the LA side, which are larger and always glabrous (Dressler 1957). Yet, plants from both fronts have evolved relatively shorter cyathia when compared to those from mainland populations, a case that was initially attributed to possible shared ancestry (Dressler 1957), but current genetic data and phylogenetic analyses support as a case of evolutionary convergence following divergence (Dressler 1957; Cacho et al. 2019). Genetic, phylogenetic, and population genetic analyses both based on a few markers (Cacho and Baum 2012) as well as at a genomic scale (Medina-Rodríguez 2022) support limited gene flow between populations of the GA and LA forms. This happens despite contrasting forms occurring in sufficiently close proximity for hummingbirds (hypothesized pollinators) to make their way between populations (Caribbean hummingbirds are known to move across islands, albeit not frequently; Lack 1973; Pulich 1968; Stimson 1944), and also for vegetative dispersal (Dressler 1957; N. Ivalú Cacho pers. obs.). Factors with the potential of driving divergence have not been studied in this system. In this work, we evaluate whether shifts in floral timing or in potential pollinators could be acting as such.

In plants, pre-zygotic factors such as shifts in floral phenology (Stiles 1975; Nuismer and Cunningham 2005; Brandenburg et al. 2009) and pollinator preferences (Fulton and Hodges 1999; Maad and Nilsson 2004) are among some of the most important barriers promoting isolation among lineages (Coyne and Orr 2004; Wiens 2004). Differences in floral phenology can be effective in preventing breeding, and because they act early on, they can have a very strong contribution to reproductive isolation between diverging plant lineages at different evolutionary scales (Coyne and Orr 2004; Martin and Willis 2007; Martin et al. 2007; Osborne et al. 2020). In a population of mixed diploid and tetraploid Chamerion angustifoilum (Onagraceae), opportunities for inter-cytotype mating were drastically reduced (by 47%) due to the combined effects of flowering asynchrony and insect foraging behaviour contributing to prezygotic mating isolation among cytotypes (Husband and Schemske 2000). Population level studies of local adaptation involving floral asynchrony show that floral phenology can evolve quite rapidly (Mitchell-Olds and Schmitt 2006; Franks et al. 2007), and it has been suggested that further divergence can lead to initial phenological differences to be strengthened by divergent selection (Chapurlat et al. 2020). In two closely related Gymnadenia orchid species that have been shown to introgress, flowering time and scent emission are under different selection regimes (Joffard et al. 2022), a scenario that is consistent with floral asynchrony playing an important role in maintaining species boundaries between closely related species (Chapurlat et al. 2020). Flowering time is under selection and phenology shifts mediate prezygotic reproductive isolation in Metrosideros nervulosa and M. sclerocarpa (Myrtaceae), two sister species that occur in sympatry (Osborne et al. 2020). At a broader scale, a study of plants representing nine genera (six families) where adaptation to harsh serpentine soils has evolved argues that phenological isolation plays a significant role across stages of adaptive divergence across clades (Sianta and Kay 2021), suggesting that these mechanisms operate more generally across plants.

The importance of pollinators in the generation and maintenance of plant diversity has long been recognized, Darwin being one of the first proponents of this idea (Darwin 1877; Stebbins 1970; Gentry 1982; Van der Niet and Johnson 2012). When interactions between plants and their pollinators are specialized, plants can present a set of traits that function in attraction and in improving efficiency of visitation and pollen transfer. As many as one fourth of the divergence events across angiosperms can be attributed to pollinator shifts (Van der Niet and Johnson 2012). At shallower phylogenetic scales, now classic studies (Schemske and Bradshaw 1999) as well as more recent ones (Temeles et al. 2016) continue to show that visitation patterns that influence pollen transfer can mediate reproductive isolation in plants. Experimental approaches suggest that changes in plant traits and mating systems (which would promote reproductive isolation) can rapidly follow changes in communities of pollinators (Gervasi and Schiestl 2017).

Reward offered by plants to their potential pollinators comes in the form of nectar or pollen, and its importance mediating interactions of plants with their pollinators is well established (Armbruster 1993; Goldblatt and Manning 2006), to the point that reward has been used as a proxy to predict a plant’s main group of pollinators (Baker and Baker 1990; Fenster et al. 2004; Zhao et al. 2016). The evolutionary ecology of pollination reward is complex, with factors as diverse as resources available to the plants, climatic fluctuations, and biotic interactions influencing reward simultaneously (Abrahamczyk et al. 2017). Yet one of the most consistent patterns substantiates the expectation that the energetic content of the reward offered by a plant is correlated to the energetic needs of its main pollinators (Heinrich and Raven 1972). There is still much to learn about the evolutionary ecology of nectar (Parachnowitsch et al. 2019), but many have successfully documented correlations between pollination systems and nectar characteristics (Perret et al. 2001; Nicolson 2002; Goldblatt and Manning 2006; Martén-Rodríguez et al. 2009; Bruneau 2014; Tavares et al. 2016; Tiedge and Lohaus 2017). In a phylogenetically informed study in Iochroma (Solanaceae) where reward, other floral traits, and visitation were quantified from observation and measurements in the field, reward was one the two attributes (the other being floral display) with direct association with pollinator preference, more so than traditionally invoked floral size and colour: plant species that produced larger quantities of nectar and larger displays were significantly more likely to be pollinated by hummingbirds than by insects (Smith et al. 2008a).

Here, we use a combination of collections-based and field-based observations and measurements to evaluate if there is evidence supporting that shifts in phenology or in floral visitors (a proxy for potential pollinators) could be acting as potential factors promoting or maintaining differentiation between the incipient lineages (GA and LA fronts) of the slipper spurge E. tithymaloides L., a plant undergoing active evolutionary divergence in the Caribbean.

All data accompanying this paper, including phenology, reward, and visitation data are deposited in Dryad (Cacho et al. 2024).

We documented phenology, reward, and visitation for Euphorbia tithymaloides across its range in the Caribbean (with emphasis in its distribution in the Antilles) and evaluated whether divergence along two fronts (Lesser and Greater Antilles) within this species is accompanied by shifts in phenology or shifts in pollinator regimes between these two incipient lineages. Through observation and measurements in natural populations and assembling data from collections (herbaria and a common garden), databases (GBIF), and online citizen science databases (iNaturalist), we document that throughout its range, E. tithymaloides flowers between the end of October and the end of May and produces (per floral unit) on average ~2.9 µL of nectar with a mean sugar concentration of 46.5 ºBrix, and a mean net sugar content of ~1.4 mg. Hummingbirds, other birds, and insects (in that order of importance) visit the inflorescences of E. tithymaloides, with hummingbirds consistently being the most important floral visitor throughout the plant’s geographic range in the Caribbean.

Other authors have shown that floral phenology can be a factor of critical importance in maintaining or promoting reproductive isolation between closely related species or among diverging populations of flowering plants across a wide phylogenetic breadth, including examples in the evening primrose family (Chamerion, Onagraceae) (Husband and Schemske 2000), orchids (Gymnadenia, Orchidaceae) (Chapurlat et al. 2020), and the myrtle family (Metrosideros, Myrtaceae) (Osborne et al. 2020), among others. Because phenological barriers act early on in preventing reproduction, they are believed to be of relevance in instances where the process of lineage divergence leads to speciation (Coyne and Orr 2004; Martin and Willis 2007; Martin et al. 2007; Osborne et al. 2020). In E. tithymaloides, our results reveal a substantial overlap (above 71%) in floral phenology among plants of from the Mainland, and the Lesser Antilles or the Greater Antilles, and above 82% for plants of the Greater and Lesser Antilles (the most distantly related in this system). The relative contribution of overlap in flowering time and vegetative reproduction (Dressler 1957) to gene exchange between the diverging lineages of E. tithymaloides (Cacho and Baum 2012; Cacho et al. 2019; Medina-Rodríguez 2022) remains to be assessed.

Environmental factors, including amount of light, rainfall, competition, and herbivory can contribute to variation in floral phenology (Elzinga et al. 2007; Matthews and Mazer 2016) and it has been suggested that when growing seasons contract or expand, selection could lead to changes in floral phenology ultimately due to climatic differences (Franks et al. 2007), which in turn could alter biotic interactions. To this effect, our data reveal a meaningful overlap in floral phenology between plants of E. tithymaloides growing in the Greater and Lesser Antilles, and this is consistent with the lack of pronounced climatic differences among these areas in the Caribbean (Cacho et al. 2019; Jacobo-Arteaga et al. 2022).

In plants, both environmental and endogenous factors (including genetic and hormonal responses) must be integrated in the control of flowering–the transition from vegetative to reproductive development (Koornneef et al. 1998; Putterill et al. 2004). While we see subtle differences in timing between plants growing in natural populations and those in our common garden, overall, the pattern is one of agreement, both in the timing of flowering activity (end of October to end of May) and in the overlap among focal areas. These results suggest that flowering in E. tithymaloides is genetically induced by short days, as others have observed (Dressler 1957). A strong genetic component in the control of floral phenological activity has been documented in several systems including both long-day plants like Arabidopsis (Johansson and Staiger 2015), and short-day ones like chrysanthemums (Higuchi et al. 2013).

The study of floral reward is complex and multivariate (Abrahamczyk et al. 2017); one of the few patterns on which there is agreement is a correlation between the energetic content of nectar and the energetic requirements of potential pollinators (Smith et al. 2008b). Accordingly, high nectar volumes and ultimately high sugar contents are observed in plants that are pollinated by hummingbirds, who are mainly nectar feeders with high energetic demands (Baker 1975; Bolten and Feinsinger 1978; Lotz and Schondube 2005; Johnson and Nicolson 2008). Also, reward has been shown to be correlated with floral display. In Iochroma (Solanaceae), species with larger displays offer rewards higher in energy and had greater probabilities of being pollinated by hummingbirds (who are visual searchers) than those with smaller display sizes, which produced a lower reward nectar and were more likely to be pollinated by insects (Smith et al. 2008a). We did not quantify floral display in E. tithymaloides, and it remains a question if differences in display have evolved among populations or between diverging lineages in this system. Yet, we do not find evidence supporting differences in amount or energetic content of nectar being offered by plants of E. tithymaloides from the Greater and Lesser Antilles (or those of the mainland).

Hummingbirds visit E. tithymaloides most frequently. They approach inflorescences in active flight and make contact with exposed anthers and stigmas with their bill when probing the inflorescence searching for nectar while hovering (Fig. 6). Other birds, mainly individuals of Coereba flaveola (Thraupidae), also visit and probe the cyathia of E. tithymaloides, but they need to perch to probe inflorescences, and thus approach them from a variety of angles several of which do not render a reliable contact of the bird’s body (bill, chin, or forehead) with the reproductive parts of the plant. Insects (hymenopterans and lepidopterans to a lesser extent) account for the minority of visits in E. tithymaloides, and from our observations, most of these visits do not translate into contact between the insect’s body and anthers and stigma of the plant: insects, while searching for nectar, land and walk on the cyathium mostly on the posterior part, where openings for the nectar chamber can be accessed for a small animal, and only rarely contact is made with its anthers or style. Altogether, these observations and data make it unlikely that insects could be playing a role as reliable and important pollinators in this system.

Our data points to hummingbirds as the main floral visitors in E. tithymaloides but we cannot exclude the possibility of a mixed pollination system, which has been documented in other plants. In Aechmea bromeliads, the combination of effectiveness of pollen transfer and frequency of visitation translates into a bimodal pollination where bees are secondary to hummingbirds in importance as pollinators, although debate exists as to whether bimodal pollination is a stable state (Schmid et al. 2011). A limitation of our study is that we did not quantify effectiveness in pollen transfer among floral visitors of E. tithymaloides, but several lines of evidence point to hummingbirds as the main pollinators. Hummingbirds were overwhelmingly more important than any other functional group as floral visitors across areas for this Caribbean euphorb. Our visitation data and field observations (registered mostly in the Antilles, particularly in the Greater Antilles) related to behaviour of floral visitors revealed that in contrast to hummingbirds, bees and other birds rarely contact floral reproductive parts (stamens or styles) of E. tithymaloides and thus support hummingbirds as the most likely effective pollinators of this plant throughout the Caribbean, a finding that agrees with what others have observed (Dressler 1957; Veiga-Blanco et al. 2013). Also, E. tithymaloides fits the pattern of plants specialized in hummingbird pollination, which tend to produce tubular-shaped flowers with red colouration (Temeles and Kress 2003; Iles et al. 2017). Moreover, hummingbirds exhibit territorial behaviour towards E. tithymaloides: individuals chase off other hummingbirds or other birds from plants or patches of plants that they consider their territory. Territoriality has been well documented in hummingbirds as one of their main foraging strategies (Sargent et al. 2021), and we observed this behaviour at many of our study sites: individuals of larger species of hummingbirds (or males within a species) would attack others (both smaller hummingbirds or other birds) and minimize their visitation but not fully prevent it. We did not observe a single instance of hummingbirds chasing off insects, a behaviour that has been documented (Boyden 1978), with more aggressive chasing the larger the insects (as would be predicted by energetic theory: other visitors should be chased accordingly to the threat they pose in terms of nectar depletion or territory displacement). An additional limitation of our study is that we observed floral visitation dynamics in a relatively small window of time (~13 h/study site) and therefore does not account for potential resource partitioning among individuals or species of hummingbird, which could take place and quickly change (Hixon et al. 1983). Floral visitor dynamics can also vary over time, space, and with community structure and composition (of plants and their interactors), among other factors (Leimberger et al. 2022). In fact, high temporal species turnover in plant-pollinator network dynamics has been documented in different systems at the community level (Alarcon et al. 2008; Petanidou et al. 2008). Yet, both plant and animal communities in insular environments tend to be less complex than their continental counterparts (Lack 1973; Kodric-Brown et al. 1984; Brown and Browers 1985), which could explain that we see no indication of obvious spatial or temporal variation in functional groups across sites. Thus, we conclude from our visitation data and observation that (albeit our modest window of observation), no evidence points to insects being important nectar feeders or potential pollinators of Caribbean E. tithymaloides. Likewise, there is no evidence of shifts in pollinator functional groups across geographic areas where E. tithymaloides occurs in the Caribbean. Yet, we cannot rule out the possibility of other visitors, including other birds (mainly Coereba flaveola) or animals in other functional groups (mainly bees) playing a role as secondary or incidental pollinators in this plant.

Other factors that could be playing an important role in contributing to divergence in E. tithymaloides rely not on a shift of pollinator functional group, but rather in changes in the identity of pollinators within the same functional group: different hummingbird species could be acting as main pollinators in different islands. Our network analyses of floral visitors of E. tithymaloides reveal relatively little overlap of hummingbird species among areas and islands/countries (as expected given their geographic ranges). Lower values of connectance and higher values of diversity and specialization in networks by islands/countries are consistent with a pattern of species replacement, especially when considering hummingbird species. Hummingbird communities in the Caribbean are complex and structure according to island size and topographic complexity (Lack 1973; Kodric-Brown et al. 1984; Brown and Browers 1985), so that different species of hummingbirds could be imposing different selection regimes on floral morphology, limiting pollen transfer, and thus driving evolutionary divergence. This scenario would be like that observed in sister species of Heliconia plants in the Caribbean, where H. bihai and H. caribaea are differentially visited by males and females of the Eulampis jugularis hummingbird, likely promoting their reproductive isolation and preventing them from collapsing into one lineage (Temeles and Kress 2003). A similar pattern has been documented in Caribbean Gesneriaceae (Martén-Rodríguez et al. 2010) and more broadly, the pattern of trait matching between a plant’s floral morphology and its hummingbird pollinators would also support this idea (Dalsgaard et al. 2021). Further investigation would be necessary to test whether such a scenario is at play across populations of E. tithymaloides.

Using a combination of data directly derived from field observations and measurements together with data gathered from collections and databases, including citizen science projects, we studied the timing of floral phenology, quantified and characterized reward (nectar), and evaluated floral visitation in Euphorbia tithymaloides across the Caribbean with emphasis in the Antilles, to evaluate if the morphological, ecological, and genetic divergence observed between incipient lineages of this plant system is accompanied by shifts in floral phenology or pollination systems. Our data are consistent with E. tithymaloides being a “short-day” plant, flowering mainly during the winter (between October and May). Morphology, reward, and visitation data support hummingbirds as the most likely and main pollinators of Caribbean E. tithymaloides, with no indication of a change of pollinator functional group in any of its main areas of occurrence. Altogether, our results do not support shifts in floral phenology or in pollinator functional groups as potential barriers to gene exchange that could be contributing to divergence in E. tithymaloides. Ecological factors promoting divergence in this system in incipient speciation remain to be described; possibilities include hummingbird species replacement throughout the Caribbean or geographical patterns in hummingbird morphology or behaviour, as has been shown in other systems.