The rate of global urbanization is increasing at an alarming pace, as the human population has grown to over 7 billion people—from 1.6 billion people in the 1900s—, half of which reside in urban areas. This increase has necessitated the expansion of urban habitat and increased consumption of natural resources. While the effects of urbanization on species diversity is well-documented (increasing urbanization decreases species diversity), its effects on species interactions have been less studied. Plant-animal interactions, such as seed dispersal, may be especially sensitive to urbanization. For plants, animal-based seed dispersal depends on several aspects, including 1) the rate of interactions with seed dispersers, 2) the probability of seed dispersal from an interaction event, 3) the identity and the number of seed disperser species –especially regarding differential ability to scarify seeds through digestion, and 4) the probability of germination after seed dispersal. Urbanization may affect seed dispersal dynamics by altering the frequency of interactions and/or the identity and diversity of seed dispersers. Consequently, the probability of seed dispersal and the ability of seeds to germinate and survive after being dispersed may be negatively affected by urbanization. In this study we ask specifically: 1) Will birds visit T. radicans at a greater rate in urban or natural habitats? 2) Will the diversity of dispersers be higher in urban or natural habitats? 3) Do seeds from urban or natural sites have a greater probability of dispersal? 4) Will seeds from urban or natural habitat be more likely to germinate? To compare differences in rate of visitation and disperser diversity between urban and natural habitats, individual T. radicans plants in two urban and two natural sites were observed for interactions by birds. Dispersal probability was estimated by marking fruits with a UV fluorescent dye and estimating a proportion of dispersed seeds at the end of the season. Seeds dispersed = total fruits marked – number of recovered fruits. Germination success will be estimated by collecting defecated—thus scarified—seeds in natural and urban sites as well as collecting non-dispersed seeds (that will be treated with either water or sulfuric acid). Seeds will be cold stratified 90 days before planting in constant 28oC and 16:8 L:D conditions. Preliminary results indicate that the rate of visitation, species diversity, and probability of seed dispersal are all significantly higher in urban sites. This trend suggests that T. radicans in urban habitat may be more successful than in natural habitat, however further research is necessary to confirm this.

Introduction, Methods Most flowering plants rely on biotic pollination for seed production. However, pollination failure can still occur due to inadequate quantity or quality of pollen delivered (pollen limitation). In fact, 60% of plant populations have been shown to be pollen limited. Pollen limitation can occur due to low pollinator availability or when low quality pollen fails to fertilize the ovules. Co-flowering species share pollinators, and can influence each other's pollination by attracting a larger pollinator pool compared to when flowering alone (pollinator facilitation), resulting in greater pollen loads. On the contrary, a limited pollinator pool can lead to pollinator competition and decrease the amount of pollen received. Lastly, pollinator competition may increase self-pollination or heterospecific pollen transfer, hence increasing pollen quality limitation. Here, we evaluate the contribution of pollen quantity and quality limitation to plant reproductive success across a gradient of co-flowering community diversity. We used four co-flowering Clarkia species in the Sierra Nevada as our study system. Species bloom singly and in combination of 1-3 other Clarkia species. Styles from all species (~60/species) were collected across 25 sites varying in co-flowering diversity. 1600 styles were processed in the lab and the amount of pollen grains and pollen tubes were counted. Results, Discussion Clarkia speciosa received the most pollen grains (477 grains) and Clarkia cylindrica the least (300 grains) on average across all populations. Preliminary data also shows that pollen quantity limitation decreases in communities where all four Clarkia species co-flower compared to when they flower singly or with only one-two other species. This pattern was observed for all Clarkia species. For instance, Clarkia unguiculata alone received 465.68 (± 30.63) pollen grains on average compared to 840.65 (±56.03) when co-flowering with all three other Clarkia species. Clarkia xantiana received 41.50 (± 48.82) pollen grains when flowering alone compared to 407.8 (± 62.19) when co-flowering with all other Clarkia species. We are currently collecting data on pollen tube number to evaluate differences in pollen quality limitation. Our preliminary data suggests pollinator facilitation can be a key mechanism reducing pollen quantity limitation, promoting species co-existence, and increasing plant reproductive success in this system. However, it is still not known if higher pollen loads lead to higher fertilization (work in progress). Clarkia species can self-pollinate (low quality pollen) or receive pollen from a different Clarkia which will not lead to fertilization and thus pollen quality could still affect reproductive success and mediate co-existence in these communities.

Urbanization is increasing at a dramatic rate as the human population increases. While it is well-known that urbanization tends to decrease species diversity (i.e., biotic homogenization), it is not known how urbanization affects the frequency and efficiency of species interactions. Seed dispersal is a plant-animal interaction that depends on disperser feeding rate, disperser diversity, probability of seed dispersal and germination. How these factors are affected by urbanization however is unknown. In this study, we evaluate how urbanization alters these factors. Urban sites had 2x higher feeding rate and 3x higher number of disperser species. The probability of seed dispersal however was the same between natural and urban sites. Moreover, the probability of germination after dispersal was 20% lower in urban sites, leading to overall negative effects of urbanization on T. radicans seed dispersal. In this study we demonstrated that urbanization can affect species diversity, as well as their ecological functions.

The rate of global urbanization is increasing rapidly as the human population expands, leading to species loss and biotic homogenization. Less studied, however, is the effect of urbanization on the frequency and efficiency of species interactions. Animal-based seed dispersal interactions may be especially sensitive to urbanization because they depend on several factors: 1) the rate of seed dispersal interactions (feeding), 2) diversity of disperser species, 3) the probability of seed dispersal and 4) the probability of germination after seed dispersal. However, how urbanization disrupts species interactions, including seed-dispersal, is still poorly known. In this study, we evaluate differences in the frequency and efficiency of seed dispersal between urban and natural sites using Eastern Poison Ivy (Toxicodendron radicans) as the focal species. Individual T. radicans lianas within natural and urban sites were observed in twelve-minute intervals (total 185.8 hours) at urban and natural sites during which the number and identity of feeding avian species was recorded. A total of 9500 fruits between natural and urban sites were marked with a UV fluorescent dye. Undispersed marked fruits were recovered via seed traps to estimate probability of dispersal. Defecated fruits were collected from natural and urban sites to evaluate germination efficiency after dispersal. Feeding rate was twice as high in urban compared to natural sites (P=0.007). Additionally, seed disperser diversity was on average twice as high in urban sites and species composition was significantly different between natural and urban sites. However, probability of seed dispersal was not significantly different between urban and natural sites (P=0.3). Interestingly, germination rate was 20% higher in defecated seeds collected from natural sites compared to defecated seeds from urban sites (P=0.005). Our results suggest that while T. radicans attracts a higher number and greater diversity of seed dispersers in urban areas, overall dispersal success is the same or even greater at natural sites, as seeds have a higher chance of germinating after being consumed by dispersers at natural compared to urban sites. Species composition differences between sites may play an important role in germinability of seeds; differences in species’ feeding strategies result in differences in their ability to scarify seeds in their digestive systems, a necessary step for seeds that rely on animal seed dispersers. Urbanization can thus negatively affect seed dispersal interactions by altering the composition of disperser species. Other animal-based interactions may be similarly affected by urbanization, and thus we emphasize the need for further studies.

Thesis (PhD)--Stellenbosch University, 2013. / Pollinators play a fundamental role in floral evolution. They can exert selection on the flowers they visit in a plethora of different ways, ranging from innate floral preferences to differences in body size and shape and behavioural elements such as flower constancy and learning capacity. Since different pollinators exhibit differences in these characters, shifts between pollinating species are often considered the most likely drivers of floral diversification. While many lines of evidence support this claim, numerous angiosperms pollinated by a single species also exhibit floral variation. Throughout my thesis, I explore and investigate floral diversification in such species in the absence of pollinator shifts. In Chapter 2, I investigate variation in the preference of conspecific male and female pollinators for the floral traits of a sexually deceptive daisy that comprises distinct floral forms. I show that its pollinator exhibits gender-specific variation in floral preferences, and that some floral forms have specialized on the male pollinator. This chapter thus illustrates the importance of intraspecific variation in pollinator preference for floral diversification, an underappreciated mechanism in this field of research. The innate preferences of pollinators are likely to have a genetic basis, especially innate preferences that govern mate choice. Genetic structure within the pollinators of sexually deceptive plants, which mimic female insects to achieve pollination, may thus provide an important source of selection on the plants they pollinate. This depends on an association between genetic divergence and divergent mate preferences, and I explore this intriguing idea in Chapter 3. While pollinators associated with sexually deceptive floral forms did exhibit significant genetic structuring, male pollinators from different phylogeographic clades all exhibited preference for the same sexually deceptive floral form, thus rejecting this hypothesis. Another behavioural attribute of pollinators that may affect floral evolution, particularly in deceptive plant species, is learning ability. Studies on sexually deceptive orchids often report that male pollinators tend to avoid sexually deceptive flowers with experience. In Chapter 4, I systematically investigate learning abilities within male pollinators and the costs they suffer on sexually deceptive floral forms that vary in deceptiveness. Results reveal a positive relationship between the level of floral deceptiveness and the 4 associated mating costs that deceived males suffer. Pollinator learning, however, appears to occur only on the most deceptive floral forms, suggesting a link between the costs suffered to the occurrence of learning. In Chapter 4, I systematically investigate learning abilities within male pollinators and the costs they suffer on sexually deceptive floral forms that vary in deceptiveness. Results reveal a positive relationship between the level of floral deceptiveness and the associated mating costs that deceived males suffer. Pollinator learning, however, appears to occur only on the most deceptive floral forms, suggesting a link between the costs suffered to the occurrence of learning. In Chapter 5, I explore the importance of florivory damage in a polymorphic daisy. Studies on floral evolution often overlook the significance of florivorous visits and focus only on pollinator-mediated selection. I show that floral polymorphism is maintained by antagonistic selection exerted by pollinators and florivores on the same floral traits. Lastly, I focus on evolutionary history to explore similarity in the patterns of South African angiosperm evolution and the pollinator species used throughout my thesis. Molecular dating shows this pollinator exhibits broadly congruent evolutionary patterns to these angiosperms, indicative of a shared biogeography. Taken together, my thesis demonstrates the vast impact of floral visitors, in particular pollinating insects, on the evolution of floral form. / My research was funded by the National Research Foundation of South Africa (NRF) and personal funding was provided by a NRF Innovation scholarship and merit bursaries from the Botany and Zoology department at Stellenbosch University. A WhiteSci Travel Grant and financial support from Prof. Erik Svensson at Lund University also allowed me to present parts of my research at international conference.

The ultimate causes of bird migration are largely unknown despite more than a century of research. By studying partially migratory short-distance tropical migrants and by employing comparative methods, some difficulties in testing hypotheses for evolution of migration can be overcome. Using comparative methods I tested the evolutionary precursor hypothesis, a major hypothesis for why migration evolved in some lineages and not in others. The results of this study conflicted with many assumptions and predictions of the evolutionary precursor hypothesis. Most importantly, migratory behavior was not related to diet and habitat in simple ways. The interaction between diet and habitat, as well as consistent associations between flocking behavior and migration suggested that food variability is poorly captured by the surrogates embodied in the evolutionary precursor hypothesis. I then employed comparative methods to studying tropical altitudinal migration. Comparisons of diets and fruit preferences between species pairs showed that migrants are more frugivorous, eat a broader diversity of fruits, and have diets that more strongly resemble their preferences than do residents. Although providing evidence that food limitation plays a role in altitudinal migration, these results do not support the hypothesis that interspecific competition explains variation in migratory behavior. Next, I provided the first test of a predation-based hypothesis to explain altitudinal migration. Migrants breed at higher elevations than where they spend their non-breeding season. Thus, birds may migrate uphill to escape high nest predation risk at lower elevations. Results from this experimental study are largely consistent with this hypothesis, but anomalies between predicted and observed patterns suggest that either migration of lowland birds occurs in response to other factors, or that anthropogenic change has altered the tradeoffs involved in migratory decisions. Finally, I focus on a single migrant species and evaluate (a) two food-based hypotheses to explain the destination of migration movements, and (b) mechanisms underlying intra-specific differences in migratory strategy. Food can explain why Corapipo altera migrate uphill, but not why they migrate downhill. My data on sex bias and body condition leads to a new hypothesis explaining the complete annual cycle of this tropical migrant bird.

Previous studies of plant-pollinator interactions have relied on the use of floral visitation data. Although, this may be insufficient to fully characterize the diversity and strength of plant-pollinator interactions. By using pollen transport data (i.e. pollen on pollinators), new insights can be gained on the structure and function of plant-pollinator communities. Yet studies that characterize and compare pollen-transport with floral-visitation networks are scarce. Furthermore, the strength and frequency of plant-pollinator interactions can vary across temporal scales. Although, monthly and within-day variation in network structure has been little studied. By evaluating variation in network structure across these biologically relevant time scales we will gain a better understanding of the factors that shape plant-pollinator communities. Here, we build plant-pollinator interactions networks based on floral visitation and pollen transport data by observing, collecting and sampling pollen from floral visitors in a southern Appalachian floral community. We aim to 1) compare the congruence of plant-pollinator networks built on floral visitation and pollen transport data and 2) evaluate within season and within-day variation in plant-pollinator network structure. To assess floral visitation and to quantify pollen transport, four 1x40m transects were set up at the study site. Morning collections were conducted between 8:00 AM and 3:00 PM twice per week while afternoon collections took place once per week between 3:01 PM and 5:00 PM over 20 non-consecutive days. All flower visitors observed interacting with a flower’s reproductive structures (i.e. searching for pollen and nectar) were collected.All collected pollinators were processed for surface pollen loads by dabbing the body with a 3x3mm fuschin jelly cube. Each area of the body was dabbed three times to standardize sampling. Identification and quantification of pollen was done using a compound light microscope. Data was then analyzed with the “bipartite” package of R to create bipartite plant-pollinator networks. Procrustes analysis was used to identify differences in network structure. Preliminary results show that the structure of floral visitation and pollen transport networks are significantly different from each other (P <0.01). Pollen-transport network size is almost four times larger (496 links) compared to the floral-visitation network (109 links). Species in the pollen transport network tend to be more connected (connectance = 2.3) and have five times more links per species on average (5.22 links) than floral visitation networks (connectance = 1.1, links = 1.8). Within-season and within-day differences in network structure are currently being evaluated. Our results so far show that pollen transport networks at our study site captured 78% more unique interactions and, thus, provide more accurate network structure. Interpretation of pollen transfer versus floral visitation networks can have important implications for our understanding of community-level functions such as their resilience and stability.

Urbanization is increasing at a rapid pace with negative consequences for native biodiversity. While it is well-known that urbanization can lead to biotic homogenization (dominance of a few competitive species), effects of urbanization on ecological functions that rely on the frequency and efficiency of species interactions are less understood. Seed dispersal success depends on seed disperser feeding rate and diversity, which can affect the probability of fruit removal (i.e. seed dispersal) and germination rates. However, how these factors are affected by urbanization is unknown. In this study, we evaluate the effects of urbanization on factors that contribute to seed dispersal success (seed disperser diversity, frequency, probability of fruit removal and germination rate using Toxicodendron radicans and its seed disperser community as a model system. We found that urban sites had three times more disperser species and two times higher feeding rate compared to natural sites. However, the probability of individual fruit removal did not differ between natural and urban sites. Moreover, germination rate after dispersal was 20% lower in urban sites, leading to overall negative effects of urbanization on T. radicans seed dispersal. We propose differences in seed germination rate are driven by changes in seed disperser species composition and their differences in seed gut transit time. This in turn affects disperser species’ ability to successfully scarify seeds. Overall, our results highlight the need to evaluate urbanization's effects on functional ecological processes, in addition to biotic homogenization effects, in order better understand and mitigate its negative impacts on biodiversity.

Plants are under selection to simultaneously attract pollinators while deterring herbivores. This dilemma can lead to tradeoffs in floral traits, which are traditionally thought to be optimized for pollinators. My dissertation addresses the ecological costs and putative functional significance of nectar secondary metabolites, a paradoxical but widespread phenomenon in the angiosperms. I investigate this issue from the pollinator’s perspective using a series of controlled laboratory investigations focused primarily on the bumble bee Bombus impatiens and the nectar alkaloid gelsemine, from Gelsemium sempervirens. I begin by demonstrating that nectar enriched with the alkaloid gelsemine significantly deters visits from bumble bees at a range of natural alkaloid concentrations. However, this aversion can be mitigated by increasing the sucrose concentration such that the alkaloid-rich nectar is more rewarding than its alkaloid-free counterpart. I then demonstrate that the consumption of gelsemine-rich nectar can inhibit oocyte development and protein utilization in bees, but that this effect is limited to bees of suboptimal condition. Continuous consumption of the nectar alkaloid gelsemine also leads to a reduction in the pathogen load of bumble bees infected with Crithidia bombi, but direct interactions between the pathogen and the alkaloid have no impact on infection intensity. Gelsemine also fails to inhibit floral yeast growth, suggesting that nectar alkaloids may not be universally antimicrobial. Finally, I demonstrate that gross nectar cardenolides from the genus Asclepias are strongly correlated with gross leaf cardenolides and that the majority of individual cardenolides found in nectar are a subset of those identified in leaves. This pattern suggests that nectar cardenolides are a consequence of defense for Asclepias; however, they may not be a costly corollary because bumble bees show an overall preference for nectar cardenolides at mean concentrations. Altogether, my dissertation provides a new perspective on the role of chemical defenses against herbivores in plant-pollinator interactions.

Plants are under selection to simultaneously attract pollinators while deterring herbivores. This dilemma can lead to tradeoffs in floral traits, which are traditionally thought to be optimized for pollinators. My dissertation addresses the ecological costs and putative functional significance of nectar secondary metabolites, a paradoxical but widespread phenomenon in the angiosperms. I investigate this issue from the pollinator’s perspective using a series of controlled laboratory investigations focused primarily on the bumble bee Bombus impatiens and the nectar alkaloid gelsemine, from Gelsemium sempervirens. I begin by demonstrating that nectar enriched with the alkaloid gelsemine significantly deters visits from bumble bees at a range of natural alkaloid concentrations. However, this aversion can be mitigated by increasing the sucrose concentration such that the alkaloid-rich nectar is more rewarding than its alkaloid-free counterpart. I then demonstrate that the consumption of gelsemine-rich nectar can inhibit oocyte development and protein utilization in bees, but that this effect is limited to bees of suboptimal condition. Continuous consumption of the nectar alkaloid gelsemine also leads to a reduction in the pathogen load of bumble bees infected with Crithidia bombi, but direct interactions between the pathogen and the alkaloid have no impact on infection intensity. Gelsemine also fails to inhibit floral yeast growth, suggesting that nectar alkaloids may not be universally antimicrobial. Finally, I demonstrate that gross nectar cardenolides from the genus Asclepias are strongly correlated with gross leaf cardenolides and that the majority of individual cardenolides found in nectar are a subset of those identified in leaves. This pattern suggests that nectar cardenolides are a consequence of defense for Asclepias; however, they may not be a costly corollary because bumble bees show an overall preference for nectar cardenolides at mean concentrations. Altogether, my dissertation provides a new perspective on the role of chemical defenses against herbivores in plant-pollinator interactions.