Pollinators are ecologically important organisms threatened by human activities. While theory predicts that networks of interactions between plants and pollinators might be robust to the local extirpation of pollinator taxa, few studies have tested this hypothesis empirically. Here, we documented patterns of floral visitation to plants naturally occurring in fragments of coastal sage scrub (n = 6 study plots) and in large, continuous scrub reserves (n = 6 study plots). Using these data, we examine how changes in pollinator diversity affect the structure of plant-pollinator interaction networks.
In 2015 and 2016, we documented 1,191 distinct combinations of interacting plant and floral visitor taxa. We recorded approximately 400 arthropod species (n = 35,481 individuals) visiting 57 plant species. Compared to plant-pollinator interaction networks in reserves, networks in fragments contained fewer pollinator taxa, exhibited increased nestedness, and were more numerically dominated by the non-native western honey bees (Apis mellifera). However, counter to our predictions, networks in fragments and reserves did not differ with respect to the number of links present per species or in the degree of niche overlap among pollinators. These findings suggest that the restructuring of networks as a result of reduced pollinator diversity in fragments caused networks to gain, rather than lose, structural robustness through enhanced nestedness. As a result, the long-term functionality of plant-pollinator interaction networks in our system may remain uncompromised by levels of pollinator diversity loss documented. Our findings also indicate that changes in network structure driven by species removals may deviate from theoretical predictions, thus calling to attention the need for more data regarding how networks are disassembled in empirical systems.