Nectar metabolomes contribute to pollination syndromes

Abstract

'Pollination syndromes', where convergent floral signals reflect selection from a functional pollinator group, are often characterized by physical features, yet floral rewards such as nectar may also reflect selection from pollinators. We asked whether nectar chemistry shows evidence of convergence across functional pollinator groups, i.e. a 'chemical pollination syndrome'. We used untargeted metabolomics to compare nectar and leaf chemical profiles across 19 bee- and bird-syndrome species, focusing on Salvia spp. (Lamiaceae), selected to maximize switching events between pollination syndromes. We found that independently derived bird-syndrome nectar showed convergence on nectar traits distinct from bee-syndrome nectar, primarily driven by the composition and concentration of alkaloid profiles. We did not find evidence for 'passive leaking' of nectar compounds from leaves since metabolite abundances were uncorrelated across tissues and many nectar metabolites were not present in leaves. Nectar and leaf metabolomes were strongly decoupled from phylogenetic relationships within Salvia. These results suggest that functional pollinator groups may drive the evolution of floral reward chemistry, consistent with our 'chemical pollination syndrome' hypothesis and indicative of selection by pollinators, but we also consider alternative explanations. In addition, our results support the notion that nectar chemistry can be decoupled from that of other tissues.

Keywords: Salvia; bee; hummingbird; metabolomics; nectar chemistry; plant defense; specialized metabolites.

Fig. 1

Predictions of the three hypotheses tested in this study across leaf and nectar tissue types. Bee‐syndrome plants are denoted by blue flowers, and hummingbird‐syndrome plants are denoted by red flowers. Distinct leaf/nectar icon colors depict distinct chemical profiles. (a) H₀: Phylogenetic signal explains chemical profile for both leaves and nectar: Close relatives share chemical traits for both tissue types. (b) H₁: Passive Leaking: nectar profile mirrors conspecific leaf profile. (c) H₂: Chemical pollination syndrome: compounds are shared within pollination syndromes. These hypotheses are nonmutually exclusive, and we may see evidence of multiple processes for different subsets of metabolites. Image was created in BioRender, MacNeill, F. (2025) https://BioRender.com/4tcxx7p.

Fig. 2

Species included in the study with chemical dendrogram of nectar metabolites, and Venn diagram showing distribution of compounds across syndromes and tissues. (a) Cladogram generated for the species included in our study with topology confidence scores (bootstrap values) shown. Bee‐syndrome flowers are denoted by blue circles; hummingbird‐syndrome flowers are denoted by red triangles. Photographs of several species are shown, with bee‐syndrome plants on the left and hummingbird‐syndrome plants on the right. Yellow ticks indicate likely independent evolutionary origins of hummingbird pollination, five of which are inside Salvia and three of which are in the outgroups. All photographs were taken by Dr. Scott Zona, except for Salvia ballotiflora Benth., which was taken from Wikipedia (user TDogg310; image cropped). (b) Compound tree for nectar metabolites (n = 466) where branch colors indicate chemical pathways. Bar color indicates the presence in bee‐syndrome (blue) or bird‐syndrome (red) plant nectar. Length of bars indicates the number of species that a compound occurs in. Several example chemical structures are shown around their corresponding chemical classes. Image was drawn using iTOL (Letunic & Bork, 2024). (c) Venn diagram showing the number of unique compounds detected across each syndrome and tissue type.

Fig. 3

Species similarity in nectar metabolome composition reflects pollination syndrome but not phylogeny. Nonmetric multidimensional scaling (NMDS) of chemical structural‐compositional similarity illustrates nectar metabolome similarity among (a) all study species and (b) within Salvia species mapped to a cladogram. Salvia leucantha, the species with the most intermediate morphological traits (see the Discussion section, for further details), clusters more closely with bird than bee‐syndrome chemical traits (indicated with arrow, b). Confidence ellipses indicate species classified as the hummingbird pollination syndrome occupy significantly different chemical space than those classified as the bee pollination syndrome using an analysis of similarities test with 999 permutations. Nectar metabolomes exhibit phylogenetic signal when outgroups are included (c), but no phylogenetic signal within Salvia species (d), as seen by the divergence among close relatives on NMDS 1, indicating that close relatives do not exhibit trait similarity in regard to their chemistry.

Fig. 4

Species similarity in nectar alkaloid composition reflects pollination syndrome but not phylogeny. Nonmetric multidimensional scaling of chemical structural‐compositional similarity illustrates nectar alkaloid similarity among (a) all study species and (b) within Salvia species mapped to a cladogram. Confidence ellipses indicate species classified as the hummingbird pollination syndrome occupy significantly different chemical space from those classified as the bee pollination syndrome using an analysis of similarities test with 999 permutations. Salvia leucantha, the species with the most intermediate morphological traits (see the Discussion section, for further details) clusters more closely with bird than bee‐syndrome chemical traits (indicated with arrow, b). Alkaloid profile does not track phylogenetic relationships and close relatives are not close in chemical space as seen in (c, d). (e) Heatmap of alkaloids detected in nectar, in which each row is a unique compound, labeled with the class of alkaloid (such as ‘simple indole alkaloid’) and colored by the number of species the compound is found in. On the left on the heatmap is a hierarchical dendrogram showing the similarity in occurrence and abundance between the two syndromes.

Fig. 5

Species similarity in leaf metabolome composition varies with phylogeny but not pollination syndrome. Nonmetric multidimensional scaling (NMDS) of chemical structural‐compositional similarity among (a) all study species and (b) within Salvia species mapped to a cladogram. Species that exhibit the hummingbird pollination syndrome are significantly less similar to each other than are bee‐syndrome species. Leaf metabolomes exhibit phylogenetic signal when outgroups are included (c), but no phylogenetic signal within Salvia species (d), as seen by the divergence among close relatives on NMDS 1, indicating that close relatives do not exhibit trait similarity in regard to their chemistry.