Beta diversity and specialization in plant–pollinator networks along an elevational gradient

Aim To assess whether the reduced nutritional resources available for pollinators due to plant community simplification along an elevational plant‐diversity gradient changes pollinator niche breadth and richness. Additionally, we evaluated how body size and proboscis length of pollinators shifted along the gradient, and whether these changes were related to pollinator niche breadth. Location An elevational gradient (2,350–3,520 m a.s.l.) on the oceanic high‐mountain strato‐volcano of El Teide (Tenerife, Canary Islands). Taxon Flowering plant and pollinator species. Methods We compared quantitative plant–pollinator networks along the plant‐diversity gradient. We calculated a set of niche‐based topological metrics that capture the degree of specialization, niche breadth and niche overlap. Furthermore, we obtained β‐diversity measures and the proportion of replacement and richness components. Results There was an overall decline in species richness of pollinators with increasing elevation. This decline was mainly driven by the loss of species along the elevational gradient, which conformed a nested subset pattern. The whole network showed less specialization, greater connectance and lower modularity towards the summit. At high elevations, pollinators were more generalized and less selective in their flower choice, showing a greater trophic niche breadth compared to pollinators at lower elevations. Mean body size of pollinators increased with elevation, and species body size and proboscis length were positively associated with the number of plant species visited. Main conclusions Overall, results indicated that the elevational gradient filters pollinator species, probably according to their thermal tolerance and ability to exploit a wide range of trophic resources. The finding that pollinators become more generalized and opportunistic at higher elevations is a novel result, which may have implications for new research into how ecological networks vary over environmental gradients. From an applied perspective, our results highlight the importance of considering the spatial variation of species assemblages when aiming to construct functionally reliable interaction networks along environmental gradients.