As the occurrence of drought proliferates globally due to increased aridity resulting from anthropogenic climate change , drought-induced pollen limitation as mediated by pollinators may become a common and pressing concern in ecosystems worldwide. The degree of pollen limitation partly depends on how pollinators move, and thus transfer pollen, amongst flowers within and between plants. For example, increased pollinator visitation within the same plant could increase instances of self-pollination, which may heighten inhibitory effects of drought on reproduction. But if pollinators increase rates of outcrossing or preferentially forage on plants producing higher quality pollen, then pollen limitation resulting from reduced water availability may be alleviated. Chapter 1, in part, is currently being prepared for submission for publication of the material. Gambel, Jess; Holway, David A. The dissertation author was the primary investigator and author of this material.Drought represents an increasingly prevalent abiotic stressor in terrestrial ecosystems . For plants, water stress decreases plant turgor, leaf water potential, and stomatal conductance, leading to reduced photosynthetic activity and slower cell growth . Accordingly, effects of drought can lead to diminished reproductive investment: decreased flower production , reduced flower size , reduced pollen production and viability , diminished nectar volume and sugar content , collection pot modification of floral coloration , and altered emission or composition of floral volatile organic compounds .
Changes in floral traits caused by drought may subsequently influence pollinator behavior and, in turn, plant reproductive success . The roles of different pollinators in mediating how plants respond to drought will be influenced by their behaviors. Generalist pollinators, for example, visit a diversity of plants, sometimes from many different families, for nectar and pollen, whereas specialists obtain pollen and nectar from a limited number of related floral hosts and often depend on resources from specific host plants to reproduce . Bees provide excellent examples of this continuum. Polylectic bees collect pollen from many unrelated plants, but specialist bees are restricted to pollen collection from few related species . Accordingly, specialists may be expected to exhibit higher floral constancy than generalists, which they do, on average, and this behavior contributes to the effectiveness of specialists as pollinators . However, exceptions also exist . As drought stress affects a suite of floral traits, it often results in reduced visitation by pollinators . Specialist pollinators that depend on pollen from a species or genus of plant will be unable to switch resources as a result of plant-to-plant variation in floral resource quality. Generalists, in contrast, may exhibit selectivity for high-quality floral resources given that they have a broader set of resources available to them. While generalist pollinators are considered more resilient to habitat and climatic disturbance compared to specialists , few studies have compared how specialist and generalist pollinators respond to drought conditions.
Minckley et al. , for example, found that bees specializing on creosote bush, which does not flower in drought, failed to emerge and visit creosote during a drought year; however, specialists of mesquite, which flowers independently of rainfall because of its ability to use tap roots to access groundwater, were able to remain active. Conversely, Mayer and Kuhlmann discovered that during a drought year, specialist and generalist bees of Rediviva emerged on time with the normal flowering season despite bloom being postponed by five weeks; this asynchrony resulted in the loss of pollen collection by bees.Here, we use field experiments to examine these relationships for cultivated squash pollinated by honey bees , which are generalists, and squash bees pruinosa, E. strenua, which specialize on Cucurbita to such an extant that they require pollen from these plants to reproduce. The squash system is highly appropriate for this type of investigation. Given that Cucurbita do not self-pollinate , they need insects to pollinate their flowers to produce fruit and seeds. Moreover, squash plants have separate male and female flowers that are large and short-lived ; these attributes facilitate examination of pollinator behavior and responses by plants. We test three predictions concerning how specialists and generalists respond to plants experiencing a gradient of drought stress and how these changes in foraging behavior may affect plant reproduction. First, generalists should exhibit selectivity while specialists should be indiscriminate in terms of displaying preferences in the quality of floral resources. Second, higher selectivity by generalists should translate into increased transfer and deposition of pollen from male flowers of non-drought stressed plants onto floral stigmas of female flowers of non-drought stressed plants . Third, increased deposition of pollen from non-drought stressed plants could increase fruit set and seed set if pollen from drought stressed plants is low quality.
In seasonally dry environments, water stress likely affects floral resource quality on a regular basis, and pollinator behavior may play an important role with respect to how those resources move amongst plant species. This study represents a novel investigation into how the effects of drought on plant reproductive success hinges on the behavior of specialist and generalist pollinators.We used a drip-line system to irrigate plants every morning with the daily amount of irrigation held constant until plants were three weeks old. After that time, when plants were mature enough to survive drought stress, we divided them into two groups. Plants either received 2.2 L water/plant/day or 0.35 L water/plant/day . Plants in the two irrigation groups were spatially interspersed with respect to one another. We used a FieldScout TDR 100© soil moisture meter to take 2 – 4 replicate measurements of volumetric water content in the soil at 20 cm below each plant one hour after watering, 2-3 times per week for the duration of the season . Volumetric water content decreased by an average of 10% for plants in the low-irrigation group compared to the high irrigation group . Mean volumetric water content, however, varied from 38 – 51% to 24 – 48% . Given that soil moisture levels exhibited substantial variation within each irrigation group, we considered soil moisture as a continuous variable in most statistical analyses but use treatment group designations in the organization of the experiment. Within the high-irrigation and low-irrigation groups, plants were randomly assigned to either a bee-pollination group or a hand-pollination group . Plants in each pollination group were spatially interspersed. Hand pollination was used to control the source of pollen and to examine the subsequent effects in terms of plant reproduction. Once squash plants began to flower in mid-July, flowers on hand-pollinated plants were bagged prior to opening to prevent bee visitation . For bee-pollinated plants, we measured traits on a subset of female and male flowers that we bagged before they opened to prevent bee visitation . On the morning that bagged flowers opened, we detached them from the plants while still bagged and then carefully removed each flower from its bag. On each flower, we measured corolla width, nectar volume, nectar concentration, pollen mass , and pollen viability . Appendix 2.1 describes methods used to measure each floral trait. Throughout the season, we kept daily tallies of the number of male and female flowers open on each plant to assess how soil moisture variation affected plant-level, seasonal flower production. For bee pollinated plants we video-recorded flowers to quantify pollinator behavior and frequency of visitation, 10 plastic plant pots and used fluorescent pigments to investigate how bees transport pollen among plants grown under different levels of soil moisture. We recorded bee visitation on high-irrigation and low-irrigation plants concurrently; therefore, each day bees could move freely between plants in each irrigation group.
Videos of male flowers and female flowers were always recorded between 0700 – 0900. For every video, we set the video camera close enough to each flower to fill the frame of the video but far enough away such that bee visitation did not seem to be affected. For each video of a female flower, we counted the number of bee visits per unit time, and for each visit measured the time that each focal bee spent in contact with the stigma and drank nectar. For each video of a male flower, we counted the number of bee visits per unit time, and for each visit measured the time that each focal bee contacted the anthers, collected pollen, and drank nectar. We also conducted plot-level surveys of bee visitation to all open flowers each day . We used powdered DayGlo© fluorescent pigments to study patterns of pollen movement by bees . We used flat toothpicks to apply pigments to the anthers of all open male flowers prior to bee access to flowers . Male flowers in the low-irrigation group received one pigment color, whereas male flowers in the high-irrigation group received the other pigment color. Color assignments were switched daily. On the stigmas of bee-pollinated female flowers, pigment particle counts were positively correlated with pollen counts . Therefore, the number of pigment particles can serve as a proxy for the amount of pollen moved by bees from the anthers of male flowers to the stigmas of female flowers and also reveal patterns of pollen movement among plants in the two irrigation groups. We measured pigment deposition on the stigmas of 56 bee-pollinated plants. For plants with multiple stigma samples , we used mean values as data points. To hand-pollinate female flowers, we first applied fluorescent pigments to two male flowers as described in the previous paragraph and then removed pigment and pollen from the anthers of these flowers. We combined the pollen-pigment mixture from the two flowers in a petri dish and used a cotton swab to generously apply this mixture to the entire stigma of each female flower . After hand-pollination, we rebagged flowers with breathable mesh bags to exclude pollinators while simultaneously permitting flowers to experience similar conditions as those experienced by flowers open to bees . See Appendix 2.2 for additional information on pollen and pigment deposition on hand-pollinated stigmas. We allowed pollen to germinate on bee and hand-pollinated stigmas for 24 h before collecting stigmas from female flowers setting fruit. This amount of time allowed for pollen grains to fertilize ovules . We stored refrigerated stigmas in 100% ethanol , and later used a dissecting microscope at 40x magnification to count the number of different colored pigment particles deposited on the stigmas . We then dyed the stigmas with basic fuchsin solution and used a dissecting microscope at 50x magnification to count the number of deposited pollen grains . We also counted stained grains present in the ethanol solution to obtain an estimate of the total pollen deposited on the stigma. We estimated pollen deposition on 48 bee-pollinated plants; for plants with multiple stigma samples, we used mean values as data points. Fifty days after pollination , we harvested mature fruits from 47 bee-pollinated plants and 19 hand-pollinated plants. Seeds were removed from fruits in the lab, and then dried, counted, and weighed. We used total seed weight per fruit to estimate seed set. For plants with multiple fruits, we used mean values as data points. We used R version 3.6.1 for all data analysis and used the package ggplot2 to prepare figures . For all general linear models, we inspected q-q plots to test for normality, used Bartlett tests to assess homogeneity of variances, and tested the residuals of each model with the Shapiro-Wilk test for normality. We used simple linear regressions to assess how soil moisture affected the floral traits listed in Table 2.1. For plants in which more than one male or female flower was measured, we used mean values as data points. For the analyses of floral traits, the following response variables were right-skewed and thuslog10 transformed to improve normality of the residuals prior to analysis: male flowers produced, female flowers produced, male flower size, and nectar volume in male flowers. The proportion of viable pollen was left-skewed and thus square transformed to improve normality. To examine how bee visitation and behavior changes with respect to plant soil moisture , we used separate general linear models for honey bees and squash bees with soil moisture and total flowers available in the plot as fixed factors. We added flower availability to the model because the number of flowers available to pollinators increased with plant soil moisture . For analyses of bee visitation to male flowers, see Appendix 2.2 for information on how variables were transformed to improve normality. For the analysis of visitation to female flowers, Eucera visitation rate was right-skewed and thus arcsine square-root transformed to improve normality of the residuals.