Plant-Pollinator Interactions in a Changing World: Cryptic Specialization, Pollinator Movement, and Landscape Genetics of Pollinator-Dependent Plants

Plant-pollinator mutualisms are one of the most prevalent and economically important mutualisms in nature. Like many other ecological systems, plant-pollinator communities are threatened by anthropogenic activity, both directly (e.g., habitat conversion and fragmentation) and indirectly (e.g., climate change). While we are aware of many of the activities that adversely impact these systems, further research is needed before we can predict how stable plant-pollinator communities are in the face disturbances. In my dissertation, I focused on: 1) tests for cryptic pollinator specialization in plants of the Heliconiaceae that could mislead our predictions about the stability of the plant-pollinator interactions involving cryptic specialists; and 2) how pollinator foraging movements relate to landscape characteristics and whether those relationships are reflected in the genetic structure of plant populations. One important component for predicting the stability of plant-pollinator communities is the degree of partner specialization in the community. Generalist strategies, where pollinators forage at the flowers of many different plant species, each of which may be visited and pollinated by many pollinator species, result in partner redundancy and reduced dependency on any one mutualistic partner. Empirical estimates of specialization are usually constructed using observations of pollinators foraging at flowers; however, results from recent experiments with Heliconia tortuosa (Heliconiaceae) demonstrated that pollen tube germination was enhanced following visits from hummingbird pollinators whose bills match the shape of the flower but not visits from mismatched pollinators. Thus, despite perceived partner redundancy, H. tortuosa may be susceptible to coextinction following local extinctions of morphologically specialized pollinator species due to a cryptic pollinator filter and therefore cryptic specialization. If the capacity for plants to cryptically filter floral visitors based on pollinator traits is widespread, this would have implications for the robustness of many plant-pollinator communities. I tested for this 'pollinator recognition' behavior in three additional Heliconia species, H. hirsuta, H. rostrata, and H. wagneriana spread widely across the Heliconiaceae phylogeny. Furthermore, I conducted experiments with H. tortuosa to test hypotheses about the mechanism of pollinator recognition and test the reproducibility of the finding that pollen tube success increases following visits from morphologically matched pollinators compared to visits from mismatched pollinators. While I found little evidence to support that H. hirsuta, H. rostrata or H. wagneriana preferentially invest in reproduction following visits by morphologically matched hummingbirds, results from my experiments are consistent with previous findings that single visits to H. tortuosa by pollen-free hummingbirds promote pollen tube success, particularly if the visiting hummingbird's bill matches the shape of the flower. However, the mechanism remains equivocal as results from my experiments did not support any of the hypothesized mechanisms of recognition. Still, successful pollen tube germination and growth in H. tortuosa appears strongly dependent on visits from morphologically matched hummingbird species that have been shown to be sensitive to forest fragmentation. Habitat fragmentation results in reductions of habitat amount in a given landscape as well as alterations to the structure of habitat within a landscape (e.g., continuous or patchy). Pollinators that are sensitive to fragmentation may abandon or fail to discover small or disconnected patches of habitat, reducing pollen flow through the landscape and potentially increasing population genetic divergence among plants persisting in isolated patches. This has implications for the genetic diversity, population persistence, and adaptive potential of plant populations in the face of environmental change. In Chapter 3, I explored how forest encroachment into alpine meadows in the Cascade Mountains, USA, could influence Rufous Hummingbird (Selasphorus rufus) pollinator movement patterns through the landscape. Using subcutaneous Passive Integrated Transponders and arrays of artificial hummingbird feeders equipped with Radio Frequency Identification data loggers, I recorded the identities of birds and the time at which they visited feeders placed throughout a mixed cover landscape in the H. J. Andrews Experimental Forest, OR, USA. Using data from four summers, I estimated the frequency of movement between food resources. The vast majority of movements were made by females which tend to be less dominant in territorial and competitive interactions. This suggests that subordinate individuals could be important to maintaining plant population connectivity. While the uncertainty in the estimates is relatively high due to few recording stations, I would expect the frequency of movements between two locations to decrease with increasing amounts of forest between them based on the fitted spacial network model. Furthermore, placing one or both feeders under coniferous forest canopy decreased the frequency of movements between the two feeders. This indicates that hummingbird foraging searches may be focused on open habitat with limited exploration into and across forested areas and has implications for pollen flow through the landscape if forest encroachment continues to shrink and fragment meadow habitat. In Chapter 4, I tested for landscape genetic signatures in a population of an ornithophilous plant species, Aquilegia formosa (Ranunculaceae), that are consistent with isolation by forest cover. While I found evidence consistent with the hypothesis that plants growing beneath taller woody vegetation are less frequently discovered and visited by pollinators, I found the opposite of the genetic signature I would predict under the hypothesis of isolation by forest cover. I proposed that this unexpected result may be due to the spatial arrangement of resources from a pollinator's perspective. In a patchy landscape, a pollinator must travel among patches of resources that are separated in space while, in a continuous landscape, a pollinator may more easily forage from one plant to the next. Because of this difference, I expect that foraging in a patchy landscape could promote more frequent long-distance pollen dispersal. The results from Chapters 3 and 4, when considered together, suggest that: 1) in the initial stages of meadow fragmentation, the frequency of long-distance pollen dispersal could increase due to the nature of foraging in a patchy landscape; 2) subordinate hummingbirds may be particularly important to maintaining connectivity of plant populations since they are more likely to be forced out of areas with prime resources and instead forage in patchy, marginal habitat; and 3) in the later stages of forest encroachment as more plants are overgrown by taller woody vegetation, fewer plants will be discovered and visited by hummingbird pollinators. These plants should contribute less pollen to the pollen pool, potentially resulting in a decrease in effective population sizes. Plant-pollinator communities are threatened on many fronts in this time of global change. My work adds to the body of scientific literature that aims to understand how plant-pollinator interactions may be disrupted by global change. In particular, I provide data lending insight into pollinator foraging behavior in fragmented landscapes and the implications for pollinator dependent plants as well as data on cryptic pollinator specialization in plants (or lack thereof). Furthermore, my work provides examples of applications of useful statistical methods (namely, spatial network models) that I believe are underutilized in ecology and evolution.