DISCUSSION
The greater amount of conspecific pollen on pollinators from urban areas suggests those pollinators exhibit greater short-term foraging specialization than pollinators in natural sites. This finding was corroborated by the higher species richness and diversity of pollen on pollinators from natural sites, suggesting that pollinators in natural sites visit more plant species during foraging trips. Finding greater specialization among urban pollinators was contrary to our hypothesis based on ecological theory that greater interspecific competition leads to greater species-level specialization in resource use (Lawlor & Smith, 1976). As expected, urban sites had lower insect species richness, but this did not translate into lower specialization. On the contrary, we found no relationship between insect species richness and specialization. A possible reason for the discrepancy with theory is that the lower specialization we found in natural areas could result from effects of species richness on interspecific competition being mediated by pollinator abundance. Total abundance of pollinator species in our study sites is unknown, but it is plausible that even in cases where species richness is high, low species evenness could prevent interspecific competition from driving short term specialization.
Among the possible drivers of short-term specialization examined in this study, we found several that may have acted synergistically to cause urban pollinators to carry higher amounts of conspecific pollen than pollinators from natural sites. First, total floral abundance of the plant community was positively associated with higher conspecific pollen amount, and the number of flowers per plant species was greater in urban areas. Since there was no difference in plant species richness between site types, it follows that pollinators would carry more conspecific pollen in urban areas. Why might total floral abundance at the site level predict conspecific pollen amount on individual pollinators? Most pollinators do not have fixed affinities for certain plants (Waser et al., 1996) and will continue to forage on flowers of particular species when those flowers are sufficiently rewarding that travel costs incurred by passing up flowers of less abundant species are low (Heinrich, 1979; Waser, 1986). In addition, pollinators spend longer on flowers if there is more, or higher quality nectar present (Thomson, 1986), which might facilitate the accumulation of more pollen. In contrast, pollinators are more likely to switch plant species when they consistently encounter flowers of a particular species that are rewardless (GrĂ¼ter et al., 2011). It is possible that pollinators experience rewarding flowers more often than pollinators at natural sites because of the higher number of flowers in urban sites. It is also possible that floral reward is greater in urban sites due to an overall lower abundance of pollinators, or more favorable ecological conditions such as moisture levels that promote production of more flowers or flowers with greater rewards. Moisture levels are higher in the urban sites used in this study (Tables S2-S4; Figure S4), so investigating whether the higher water availability translates into higher floral reward in urban sites would thus be worthwhile.
Second, common pollinators were more specialized at urban sites than at natural sites. Relative to natural sites, average conspecific pollen amounts increased by 54% and 23% for A. mellifera and B. vosnesenskii in urban sites. The increase in specialization of these common species coupled with their greater abundance at urban sites likely contributed to the higher conspecific pollen amounts found on pollinators at urban sites. Third, pollinators carried more conspecific pollen when caught on invasive plants, and floral abundance of invasive plants was higher than that of non-invasive plants in urban areas. Therefore, the preferential foraging on flowers of invasive species likely led to the greater conspecific pollen amounts on pollinators in urban sites. The high amounts of conspecific pollen on pollinators foraging on invasive plants suggests that pollinators exhibit high fidelity to invasive plant species in our urban sites. As for all plant species, pollinator preference for invasive plants is context specific, and tends to depend on the size of floral display, quality of food resources, or plant density (Brown et al., 2002; King & Sargent, 2012). The most common invasive species across sites in our data set were Wild Radish (Raphanus raphanistrum ), Wild Mustard (Sinapis arvensis ), and Hairy Vetch (Vicia villosa ). Each of these tend to be present at high densities and have high amounts of nectar or pollen reward (Davis et al., 1998; Marchand et al., 2015; Sahli & Conner, 2007). Our finding that pollinators have higher floral fidelity to invasive species motivates future research into how floral reward or plant density may differ between native and invasive plants and how these differences may influence pollinator choices in urban environments.
Another potential driver that may contribute to the higher specialization we observed for urban pollinators is the amount of energy pollinators may need to spend acquiring sufficient resources in different environments. The importance of pollinator movement to the acquisition of sufficient resources is largely unknown (Harrison & Winfree, 2015), but pollinators have been observed to spend longer amounts of time in urban flower patches than in large continuous countryside populations (Andrieu et al., 2009). Urban landscapes are characterized by large regions of inhospitable habitat over which pollinators may need to travel to reach food resources. This longer time spent in urban fragments presumably allows pollinators to recoup energetic costs of travel among patches. After arriving in an urban patch, pollinators may be likely to continue foraging on the same species of plant and not expend energy learning to manipulate alternative floral types. The extent to which pollinators move among urban fragments in our study area would be a valuable future research direction.
Our findings have implications for plant species coexistence both in urban and natural areas. On the one hand, the negative frequency dependence in pollinator preference suggests a mechanism by which plant species may coexist with one another and persist in urban fragments. On the other hand, pollinator preference for invasive species may facilitate invasive plant spread and loss of rare native plant species. Many studies have reported positive relationships between urbanization and invasive plant abundance (Bradley & Mustard, 2006; George et al., 2009; Seabloom et al., 2006). Our results highlight a mechanism that may underlie this pattern, whereby pollinator fidelity to invasive plants facilitates their persistence in urban environments. Our results also suggest that pollinator foraging choices may facilitate invasive plant spread in natural areas. Pollinators were even more specialized when they were caught foraging on invasive plants in natural environments than they were when caught foraging on invasive plants in urban environments. The fidelity to invasive plants, in both urban and natural areas, suggests that rare native plant species may receive many fewer visits by pollinators than common, invasive species. This may be particularly problematic for the persistence of rarer native plants in urban areas that are present at low abundance. Urban pollinators caught on native plants still carried on average 53% conspecific pollen, but this percent dropped to 42% for native plants that were present at less than 20% frequency, and it dropped to 29% for plants that were present at a frequency of 10% at the study sites. These decreases the proportion of conspecific pollen with increasing rarity of native plants suggest that continued augmentation of urban flora, along with removal of invasive species may therefore be necessary to maintain plant species diversity at these sites.
To our knowledge this is the first study that quantifies conspecific and heterospecific pollen carriage of pollinators in natural and urban sites. The conspecific pollen proportions found on pollinators suggest there is a greater potential for conspecific pollen transfer among plants in urban areas than in natural areas. It is worth noting that we did not explicitly measure flower constancy, so we cannot know if pollinators moved among flowers of the same individual plant or among flowers of different individuals. Self-incompatible plants cannot produce seeds unless they receive pollen from a different conspecific individual (Castric & Vekemans, 2004). Therefore, it will be worthwhile to explore whether the greater conspecific pollen proportions found on urban pollinators are reflective of visits to single plant individuals or several different plants of the same species.