Many plants have evolved nutrient rewards to attract pollinators to flowers, but most research has focused on the sugar content of floral nectar resources. Concentrations of sodium in floral nectar (a micronutrient in low concentrations in nectar) can vary substantially both among and within co-occurring species. Sodium concentrations in floral nectar might play an important and underappreciated role in plant-pollinator interactions, especially because many animals, including pollinators, are sodium-limited in nature. Yet, the consequences of variation in sodium concentrations in floral nectar have gone largely unexplored. Here, we investigate whether enriching floral nectar with sodium influences the composition, diversity, and abundance of pollinator interactions. We experimentally enriched sodium concentrations in four plant species in a subalpine meadow in Colorado, USA. We found that flowers with sodium-enriched nectar received more visits from a greater diversity of pollinator visitors throughout the season. Each pollinator species foraged more frequently on flowers enriched with sodium and showed evidence of other changes to pollinator foraging behavior, including greater dietary evenness. These findings are consistent with the ‘salty nectar hypothesis,’ providing evidence for the importance of sodium limitation in pollinators and suggesting that even small nectar constituents can shape plant-pollinator interactions.

Experimental design. We selected four focal species for this experiment: Delphinium barbeyi (Ranunculaceae), Erigeron speciosus (Asteraceae), Helianthella quinquenervis (Asteraceae), and Heliomeris multiflora (Asteraceae). These plant species have varied floral morphologies and have an abundant and diverse community of floral visitors (CaraDonna and Waser 2020; Bain et al. 2022). All focal plant species co-occur in this subalpine meadow, are abundant, and co-flower (in July). For each species, we selected 5 paired plants; pairs were typically within 1 meter of one another, similarly sized, and had a similar number of flowers. We labeled D. barbeyi and E. speciosus individuals but considered adjacent groupings (2-5 individuals) of H. quinquenervis and H. multiflora a “single plant” because each individual has few or one inflorescence. For each pair, one plant received the sodium-enriched nectar treatment and the other received the control treatment. Control nectar contained 50% sucrose (weight : volume) and sodium-enriched (NaCl) experimental nectar contained 50% sucrose + 0.5% NaCl (weight : volume). To make the artificial nectar solutions, we combined 170 g of sugar and 200 ml of water for 0.5 °Bx. For the sodium-enriched treatment, we added 1.55 g NaCl (~ 0.5% NaCl w:v). The amount of artificial nectar applied to the flowers approximates naturally occurring nectar volumes and sugar concentrations observed at the Rocky Mountain Biological Laboratory (Table S1; Hiebert and Calder 1983; Luo et al. 2014; Kirschke and CaraDonna, unpublished data). Observations of sodium concentrations in floral nectar of 14 species in the Colorado Rocky Mountains range from < 0.001% Na to 0.12% Na (Hiebert and Calder 1983). Therefore, the 0.5% NaCl (0.077% Na) sodium-enriched nectar was approximately 5× naturally occurring concentrations of sodium in D. barbeyi at the RMBL (Table S1).

Sodium-enrichment and pollinator observations. We applied the experimental nectar treatments and observed interactions between flowering plants and insect pollinators on warm, sunny days. Nectar treatments were randomly assigned to one of the plants in each pair and applied by gently inserting a pipette tip into the nectar spur (D. barbeyi), into individual florets (H. quinquenervis, H. multiflora), or spread across the surface of the capitulum (E. speciosus) for all open flowers. We replaced the pipette tip between each experimental addition to avoid cross-pollination and cross-contamination of nectar solutions. Our experimental plants received the same nectar treatment for the duration of the experiment.

After applying nectar treatments to a pair of plants, we conducted 20-minute observations of pollinators at each plant in random order, recording the identity of each pollinator visitor, the total number of flowers visited, and the duration of its foraging bout. We recorded an interaction when an insect contacted the reproductive parts of an open flower and identified visitors in the field to the finest taxonomic unit possible. If species-level identification was not possible in the field, pollinators were recorded as distinct morphospecies (e.g., Bain et al. 2022). We did not destructively sample pollinators to avoid influencing interactions during other observation periods.