Methodological Effects on Plant–Hummingbird Networks: A Comparison of Pollen Load and Focal Observation Approaches in Colombian Andean Ecosystems

Abstract

Ecological interactions, such as plant–pollinator interactions, are fundamental to maintaining biodiversity and sustaining ecosystem functions. In tropical ecosystems, hummingbirds serve as key pollinators due to their specialized morphological traits, which facilitate plant–pollinator interactions. Understanding these dynamics is crucial for assessing ecosystem stability and biodiversity patterns. Two common approaches to studying plant–hummingbird interactions are focal observations of hummingbird visits to flowers and pollen load analyses from captured hummingbirds. Although these methods capture different aspects of the interaction process (behavioral visitation vs. realized pollen transport), both are widely used to infer plant–pollinator networks. However, because they may detect different subsets of interactions, it remains unclear to what extent they provide comparable representations of network structure and interaction composition. Despite their frequent use, direct comparisons of these methodologies remain scarce, especially in high-altitude ecosystems where hummingbirds are critical pollinators. Fieldwork was conducted in high Andean ecosystems in Colombia. Interactions were recorded through focal observations of hummingbird visits to flowers, complemented with pollen load analyses from captured individuals. To evaluate each method, interaction networks were constructed and compared using network metrics. Pollen load analysis detected higher hummingbird and plant species richness across sites than focal observations. Networks derived from pollen load data exhibited greater connectance and nestedness, while observation-based networks showed higher specialization and modularity. These differences suggest that each method captures different structural properties, with pollen load analysis offering broader coverage of interaction diversity. Pollen load analysis proved more effective at quantifying plant–hummingbird interactions, although both methods provide complementary insights.