Partitioning interaction turnover among alpine pollination networks: spatial, temporal, and environmental patterns

Abstract

Ecologists have taken two distinct approaches in studying the distribution and diversity of communities: a species-centric focus and an interaction-network based approach. A current frontier in community-level studies is the integration of these perspectives by investigating both simultaneously; one method for achieving this is evaluating the relative contributions of species turnover and host switching towards interaction turnover (i.e., the dissimilarity in interactions between two networks). We performed observations of plant-pollinator interactions to investigate (1) patterns in interaction turnover across spatial, temporal, and environmental gradients and (2) the relative contribution of pollinator species turnover, floral turnover, simultaneous pollinator & floral turnover, and host switching towards interaction turnover. Field work was conducted on the Beartooth Plateau, an alpine ecosystem in Montana and Wyoming, with weekly observations of plant-pollinator interactions across one growing season. Interaction turnover increased through time, with magnitudes consistently greater than 80%, even at time intervals as short as one week. Floral species turnover (41%) and simultaneous floral and pollinator species turnover (36%) accounted for almost all interaction turnover while host switching accounted for only 5%. Interaction turnover also significantly increased with spatial and elevational distance, albeit with lesser magnitudes than with temporal distance. The marginal spatial pattern was present for only some taxa (Bombus spp. and solitary bee species), potentially indicating variable habitat use by pollinators across the landscape. Weak environmental trends may be a consequence of unmeasured environmental variables, yet our finding that environmental gradients structure plant-pollinator interaction partitions had not previously been tested with empirical data. Our observations suggest that host switching does not readily occur at the scales of alpine flowering phenology (i.e., ∼1 week); however, whether lack of host switching is indicative of inflexible pollinator foraging, or, more likely, a lack of opportunity or necessity to switch hosts, requires further investigation.