Abstract

Interactions between plants and their flower visitors underpin pollination processes, yet these relationships can shift as flowering phenology and community composition change. Such temporal dynamics remain poorly documented in insular ecosystems, where species turnover is low but ecological roles may vary seasonally. We examined seasonal variation in the composition and structure of a plant–flower visitor network on a small tropical island. Using a temporally resolved network approach, we quantified visitation frequency, species specialization, the structural roles of plants and visitors, and tested whether flower size influenced visitor diversity and activity. Visitation rates and species strength were similar across early, mid, and late flowering periods, but network composition and specialization changed markedly. Visitor specialization was lowest at the beginning and end of the season and peaked mid-season, indicating increased generalization at seasonal edges that may buffer temporal mismatches and enhance network robustness. Both plants and visitors shifted structural roles through time, with a few taxa, including an endemic bee, recurrently acting as module hubs or connectors. Visitation frequency, but not visitor richness, increased with flower size throughout the season, suggesting that floral traits influence interaction intensity and the distribution of pollination services. Our findings show that apparently constant visitation patterns over the whole flowering season can mask the internal reorganization of species roles and trait-mediated interactions. Recognizing the hidden seasonal dynamics within pollination networks is therefore essential for improving conservation strategies and anticipating how small insular ecosystems may respond to increasing anthropogenic pressures and biodiversity loss.

Dataset description

Datasets on plant-flower visitor interactions collected on a small granitic inner island in the Seychelles over three consecutive months (October – December 2016), spanning the main flowering season. Aride Island (-4.2121, 55.6658; 135 m a.s.l.; 72 ha).

Early_Oct.csv , Mid_Nov.csv , Late_Dec.csv

Quantitative interaction matrices representing plant–flower visitor interactions during the early, mid-, and late-flowering season. Plant species are arranged in rows, flower visitor species in columns, and cell values indicate visitation frequency (visits flower⁻¹ h⁻¹), calculated as the number of visits per flower per hour of observation.

• Early_Oct.csv — quantitative interaction matrix for the early flowering season (October 2016).
• Mid_Nov.csv — quantitative interaction matrix for the mid flowering season (November 2016).
• Late_Dec.csv — quantitative interaction matrix for the late flowering season (December 2016).

Corolla_size.xlsx

This dataset contains plant species and corresponding corolla size information, expressed as both a quantitative measure and a qualitative size category. Corolla size corresponds to flower diameter (in centimeters) and was compiled from published literature sources.

Quantitative values represent corolla diameter in centimeters. Qualitative size categories were derived from these values using the following thresholds:

• Small: 0–2.0 cm
• Medium: >2.0–5.0 cm
• Big: >5.0 cm

Variables

• plant_species: Scientific name of the plant species.
• flower_size_quant_cm: Corolla diameter (cm).
• flower_size_cat: Categorical classification of corolla size (Small, Medium, Big), based on the thresholds defined above.