TY - JOUR
AU - Sorenson, Clyde, E
AB - Abstract Dioecy is rare among flowering plants, and is associated with a high frequency of threatened species. Dioecious plants are often pollinated by wind or insects, but are susceptible to pollination failure should male and female plants become spatially separated, or should pollinator abundance decline. Here we characterize the plant–pollinator interactions of Rhus michauxii Sarg (Sapindales: Anacardiaceae), an endangered dioecious shrub endemic to the southeastern United States. Working in the sandhills region of North Carolina, we detected a diverse community of arthropods visiting R. michauxii flowers, including 55 species or morphospecies, with moderate niche overlap between male and female flowers. Although most visitors acquired pollen from male flowers, pollen loads were greatly reduced or diluted on visitors to female flowers; conspecific pollen was completely absent at all-female sites. Bees in the genus Megachile appear to be the most important pollen vectors in this system because of their abundance and pollen load composition. We constructed a regional pollen transport network involving 73 arthropod species and 46 pollen species/morphotypes, in which R. michauxii participated in 10% of links and attracted 38% of individual visitors, suggesting that it competes successfully with other plants for visitation. Finally, time-lapse videography revealed that female inflorescences were visited about six times less often than male inflorescences, but at similar times of day. Despite overall high rates of bee visitation, pollen movement from male to female plants was uncommon, and restoration of sexual reproduction in this species may require hand pollination or translocation of suitable mates to single-sex sites. Dioecy—the breeding system involving separate male and female individuals—is nearly universal in animals, but rare among flowering plants. Globally, only 6% of Angiosperm species are dioecious (Renner and Ricklefs 1995), and in temperate zones, the number drops to 4% (Bawa 1980). Although dioecy may prevent inbreeding and improve resource-use efficiency through sexual specialization (Charlesworth 1993), its evolutionary success rate is low. Dioecious clades are less species-rich than their nondioecious sister groups, and are more likely to include threatened species, suggesting a heightened risk of extinction and greater conservation challenges for dioecious plants (Heilbuth 2000; Vamosi and Vamosi 2004, 2005) . Although the proximate causes of extinction risk in dioecious plants are poorly documented, reduced mate assurance and pollination failure are thought to play a role, particularly under conditions of anthropogenic disturbance (Pannell and Barrett 1998, Vamosi and Vamosi 2005). Meta analysis indicates that habitat fragmentation is more detrimental to reproduction in plants that cannot self-pollinate (including dioecious and other self-incompatible species) compared with those that can (Aguilar et al. 2006). For example, habitat fragmentation or artificially low population densities may separate the sexes by distances too great for pollen transfer by wind or insects—the most common pollen vectors in dioecious species (Bawa 1980). Direct evidence of failed pollen transfer in dioecious species is rare (but see House 1992, 1993); strong circumstantial evidence, such as higher fruit set in females growing near conspecific males or in male-biased populations, is more common (e.g., Mack 1997, Carlsson-Graner et al. 1998, Somanathan and Borges 2000, Kawelo et al. 2012, Brown et al. 2015). In addition to the risk of spatial separation of the sexes, animal-pollinated dioecious species are highly sensitive to declines in pollinator visitation. Male flowers often attract more and longer pollinator visits than do female flowers because the pollen itself is a reward to visitors, or because they produce more attractive secondary traits such as longer petals (Bierzychudek 1987, Ashman 2000). These systems succeed when pollinators are abundant, but if pollinator populations decline or other disturbances alter pollinator behavior, the less-preferred female flowers risk receiving too few visits, leading to strong pollen limitation of seed set (Vamosi and Otto 2002). One potential disturbance to pollinator behavior is rarity of the plant itself. Theoretical, experimental, and observational studies indicate that small or sparse populations of flowering plants can fail to attract pollinator visits (Rathcke 1983, e.g., Ågren 1996, Kunin 1997b, Metcalfe and Kunin 2006). In addition, pollinator fidelity to a given species tends to decrease at low floral densities, such that pollen of the rare species is less likely to reach conspecific stigmas (Rathcke 1983, Kunin 1997a). Thus, rare, animal-pollinated, dioecious plants face multiple disadvantages: Rarity alone may reduce pollinator visitation and pollinator effectiveness; these issues are compounded in situations where female flowers are relatively unattractive and lose flower visitors before males do. Such plants further lack any reproductive assurance mechanism should visitation cease or should the sexes become spatially separated. These risks to dioecious plants complicate their conservation and restoration (Brown et al. 2015, Orsenigo et al. 2017). Understanding which, if any, of these processes is at work in a rare, dioecious species is essential to its conservation management. Here, we present a case study of pollinator interactions in a rare, dioecious shrub, Rhus michauxii Sarg (Sapindales: Anacardiaceae), a federally endangered species endemic to the southeastern United States. Anecdotally, the species suffers from poor seed set (USFWS 2014), but the identity, behavior, and effectiveness of its pollinators have not been investigated. Thus, we characterize its flower-visitor community and identify the species most likely to deliver conspecific pollen to female flowers. Second, we compare the roles of R. michauxii male and female flowers within their local pollination network and identify coflowering plants that may compete with or facilitate R. michauxii via shared pollinators. Finally, we examine differences between male and female inflorescences in the duration, frequency, and timing of pollinator visits. Our goals are to understand existing risk factors for pollination failure in R. michauxii, and to provide information relevant to its management and recovery. Materials and Methods Study System R. michauxii is a 0.2–1 m tall shrub (Fig. 1a) that spreads clonally via adventitious roots (Braham et al. 2006). It was listed as federally endangered in 1989; of the 43 extant occurrences known in 2014, 33 were located in North Carolina, and 21 of those in the sandhills region (USFWS 2014). Previous molecular analysis of North Carolina populations showed that individual sites typically house one to seven clones (Sherman-Broyles et al. 1992). Anecdotal or unpublished evidence suggests that individual clones may depart from strict dioecy by changing sexes between years, or occasionally producing morphologically perfect flowers, although these phenomena have not been clearly documented (USFWS 2014). In our study area, we have so far observed strict dioecy with consistent sexual expression in 2016 and 2017. Fig. 1. View largeDownload slide R. michauxii female plant (a) and current range of the species (b). Map shows southeastern United States; counties occupied by R. michauxii are highlighted and the study area is circled in black. Species range data are from the U.S. Fish and Wildlife Service, background map is from naturalearthdata.com and in the public domain. Fig. 1. View largeDownload slide R. michauxii female plant (a) and current range of the species (b). Map shows southeastern United States; counties occupied by R. michauxii are highlighted and the study area is circled in black. Species range data are from the U.S. Fish and Wildlife Service, background map is from naturalearthdata.com and in the public domain. Within the sandhills region, we worked at 10 sites (Fig. 1b): three in the state-owned Sandhills Game Lands, two in Fort Bragg Military Reservation, and five in Camp Mackall Military Reservation. All were characterized by sandy to loamy soils in open, dry, longleaf pine forests subject to prescribed fire every 1–3 yr. Sites were separated by 300 m to 37.5 km. Although some flower-visiting insects can fly more than 300 m, making these sites not strictly independent, we treated them separately because they were separated by habitats unsuitable for continuous pollinator foraging (shaded or lacking flowers), and because they are identified and managed separately in the land managers’ endangered species programs. Two of the sites had all female inflorescences, five all male, and three mixed. In 2017, open flowers were first detected on June 5 and last detected on June 24 (except for a single all-female site where a few stems bloomed, individually and sporadically, until early September). This study included plants that flowered between June 7 and June 22. Additional study site characteristics are presented in Supp Text 1 and Table 1 (online only). Sampling To assess the diversity, identity, and importance of potential pollinators visiting R. michauxii flowers, we sampled flower visitors in three ways. First, we spent 13.7 person-hours collecting specifically from R. michauxii flowers. The time was divided across three sites (two mixed, one female) and four dates (7, 12, 13, and 14 June 2017). Each site received 1.7 to 5.4 person-hours of search time (excluding specimen handling). During this time, we searched sites and collected all arthropods observed to visit (feed from, rest in, or walk across) R. michauxii flowers. We collected arthropods directly into plastic vials, or netted them and transferred them into vials, then anesthetized them on ice packs in a cooler until transferring them to a freezer at the end of the day. To avoid pollen contamination, all vials were previously unused, and nets were washed with 70% ethanol after each capture. Second, to construct the pollination network of which R. michauxii is a part, we sampled arthropods from R. michauxii and coflowering species at nine sites (three mixed, one female, five male). We sampled each site once between 13 June and 22 June 2017. To guide sampling, we delineated one 50 m × 50 m plot per site, with one edge parallel to the long axis of the R. michauxii population. We placed three transects perpendicular to that edge by first dividing the edge into three equal sections and then using a random numbers table to locate the origin of one transect in each section. If all three transects failed to pass within 1 m of flowering R. michauxii, we moved the nearest transect so that the focal species would be included in the sampling. Along each transect, we surveyed flowers and flower visitors. To survey flowers, we counted the number of flowering stems of each species detected within a 1 m wide band centered on each transect (total area surveyed 150 m2 per site). For up to 10 stems per species, we recorded the number of open flowers (or, for Asteraceae, inflorescences) per stem. We also recorded but did not quantify additional species found within the 50 m × 50 m site but not on the transects. For each species detected, we collected an anther sample for later construction of a pollen reference library. Plants were identified according to Weakley (2015) and Sorrie (2011). To sample flower visitors, we spent 30 person-minutes (excluding specimen handling) per transect, slowly pacing the transect and capturing all arthropods visiting flowers within 2 m. To detect interactions involving locally uncommon, off-transect plants, we also spent 30 person-minutes searching the rest of the 50 m × 50 m site, for a final total search time of 2 h per site (18 h across all sites). We chose this approach, rather than timed observation of individual plants or quadrats, because visitation to R. michauxii ceased after a few minutes of sampling, then resumed after the observer moved away. We avoided this effect by moving continuously along the transects. Timekeeping and specimen handling were as described earlier for targeted sampling of R. michauxii. Finally, whereas the targeted and network sampling provided brief snapshots of the flower-visitor community across many plants at each site, we also used time-lapse cameras to obtain a longer-term record of visitation to a limited number of plants. We deployed time-lapse cameras in weather-resistant housings (TLC200 Pro camera, ATH120 housing, Brinno, Taipei City, Taiwan), with each camera focussed on a single inflorescence. Cameras took an image every 2 s during daylight hours, automatically stopping in low light (Edwards et al. 2015). Plants subject to videography were not randomly selected. Instead, we chose plants that were early enough in their phenology to provide at least two full days of video; were located where a tripod could be securely staked; and were not visible from nearby roads. We obtained scorable footage for 12 plants at eight sites, as follows. At three mixed sites, we had footage for one male and one female plant per site. The other sites had one plant each, except for an all-male site where total video observation time was curtailed by fire. Here we pooled footage for two plants into a single sample. Sample and Video Processing In the lab, we swabbed each captured flower visitor with a small cube (8 mm3) of fuchsin gel to remove and stain pollen grains, then melted the gel onto a microscope slide and sealed it with a cover slip (Kearns and Inouye 1993). For bees with specialized pollen-carrying structures, we did not swab those structures because packed or moistened pollen is less available for pollination. To prevent contamination between samples, we cleaned tools and work surfaces with 70% ethanol after each specimen. We constructed a pollen reference library using the same methods to sample, stain, and slide-mount pollen from anthers of identified plants. After sampling pollen from flower visitors, they were pinned, labeled, and identified to species or morphospecies using Fall (1899), Liljeblad (1945), Enns (1955), Bradley (1957), Mitchell (1960), Linsley and Chemsak (1961), Painter (1962), Howden (1968), Bohart and Menke (1976), Bohart and Kimsey (1980), McAlpine (1981), Bohart (1982), Stehr (1987), Thompson et al. (1990), Arnett et al. (2002), Hall and Evenhuis (2003), Ubick et al. (2005), Kits et al. (2008), Gibbs (2011), Miranda et al. (2013), Ascher and Pickering (2014); and comparison to specimens in the NC State University Insect Museum, where voucher specimens will be deposited. Difficult specimens were identified or confirmed by experts. To assess pollen load composition, we used a compound light microscope with graduated mechanical stage to scan all pollen slides from flower visitors at 100× magnification, viewing the entire stained sample in nonoverlapping, 2 mm-wide transects, increasing magnification to 400× as needed to identify individual grains. For slides from arthropods collected during targeted sampling, we noted the presence or absence of R. michauxii pollen and other pollen; when R. michauxii pollen was present, we counted the number of grains of R. michauxii and other pollen, but did not identify the other pollen. For slides from arthropods collected during network sampling, we counted and attempted to identify all pollen grains on each slide. R. michauxii pollen was distinct from that of other species flowering at our sites except for Rhus × ashei hybrids, which coflowered with R. michauxii at one site in early June. Samples collected at this site during the period of Rhus × ashei flowering are not included in this study, so we assume that all Rhus pollen in the samples presented here represents R. michauxii. Male Rhus glabra was not found flowering at or near our sites, and Rhus copallinum only flowered after the conclusion of the study. Pollens that did not match the reference library were designated as unknowns, and pollens of some species were pooled into morphotypes (Lespedeza virginica group, Lespedeza bicolor group, Galactia mollis group) because they could not be reliably distinguished at 400× magnification. We reviewed time-lapse footage using the frame-by-frame controls of QuickTime Player v7.7.6 (Apple, Cupertino, CA). We scored two full days of video (28 daylight hours, 6:00–19:59:59) per plant, except for the site where video was curtailed by fire; here we pooled video for two plants, each monitored for 20 daylight hours. During each daylight hour, we recorded the identity (genus, species, or morphotype) of each visitor and the frames in which it first and last contacted the inflorescence before disappearing. We did not score visits by spiders or ants because they are poor pollen vectors whose behaviors made visits difficult to define. Occasionally, an inflorescence would become so busy (>5 conspecific visitors simultaneously) that it was impractical to score all individual visits. In these cases, we scored visits for 15–30 min of each hour and recorded the amount of time scored as an offset variable for analysis. Similarly, we noted any time excluded due to rain. Because the videos showed one side of the large inflorescence, visitors sometimes disappeared behind it and we could not distinguish their reappearance from a new visit; likewise, we may have missed visits that remained entirely out of view. Although this is a weakness of the method, we expect it to be a constant (unbiased) source of error across plants. Data Analysis All analyses were conducted using R v.3.3.1 within RStudio v.1.0.136 (R Core Team 2013, RStudio Team 2015). R. michauxii Flower Visitors To assess the diversity and importance of flower visitors to R. michauxii throughout the study area, we considered all arthropods collected from R. michauxii flowers during both targeted sampling and network sampling. To estimate the species richness of this assemblage, we computed the Chao1 diversity estimator using the iNEXT package; we also compared the rate of species accumulation for visitors to male and female inflorescences (Hsieh et al. 2016). Then we asked whether the probability of carrying R. michauxii pollen differed among arthropod orders, using a Fisher’s exact test implemented in the ‘stats’ package. For this analysis we considered orders with n ≥ 10 individuals (thus excluding Coleoptera, Hemiptera, Lepidoptera, and Orthoptera). We divided the order Hymenoptera into two groups, bees and nonbees, because of bees’ unique adaptations for pollen transport. Because pollination in a dioeceous system depends upon visitors arriving at female flowers with pollen previously obtained from male flowers, we examined differences between male and female plants in visitor identity and the presence and quantity of pollen they carried. First, we compared the visitor communities of male and female plants. We used EcosimR package (Gotelli et al. 2015) to compute Pianka’s O, a measure of niche overlap (Pianka 1973), for male and female flowers, and assessed its significance relative to a null model in which the visitor community composition was shuffled within male and female flowers 1,000 times while retaining the original niche breadth of each sex. Next, we used an exact contingency test to ask whether the probability of a visitor having pollen on its body depended on the sex and site type of the flower from which it was captured, as represented in the following categories: male flower, female flower in mixed site, female flower in all-female site. Because our sample size for all-female sites was small in 2017 (11 visitors), we performed a second version of this analysis including an additional 16 arthropods collected at an all-female site during pilot work in 2016 (29 June and 6 July; Supp Table 2 [online only]). Finally, given pollen presence on a visitor’s body, we also expected its quantity to depend on flower sex and visitor identity. Here we considered the subset of visitors that had R. michauxii pollen on their bodies, and selected the eight most abundant species that included at least three specimens caught on female plants. In this analysis, we treated conspecific male and female bees as separate ‘species’ because females have morphological and behavioral adaptations for pollen transport that are absent in males. We used the glm.nb function in the MASS package (Venables and Ripley 2002) to construct a generalized linear model with negative binomial error distribution and log link function, in which plant sex, visitor species, and their interaction were the predictors and number of R. michauxii pollen grains per visitor was the response. We assessed model fit using the diagnostic plots of the DHARMa package (Hartig 2018), and used the emmeans package (Lenth et al. 2018) to tabulate estimated marginal means for each visitor taxon and their pairwise significant differences using the Tukey method. Ultimately, pollinator quality for a given plant species depends not only on the abundance of the visitor and the number of conspecific pollen grains it carries, but also on its fidelity to that plant species and the efficiency with which it transfers pollen onto receptive stigmas (Primack and Silander 1975, Schemske and Horvitz 1984, Vázquez et al. 2005). To synthesize several aspects of pollinator quality into a single value, we calculated a pollinator importance (PI) index, as described by Youngsteadt et al. (2018). Here we define potential pollinators as only those specimens found at female flowers with R. michauxii pollen on their bodies. The value of the PI index for species X was calculated as PIx=relative abundance of X∗(pollen carrying capacity of X)∗fidelity of X where relative abundance is the proportion of all potential pollinators that are of species X; pollen carrying capacity is the mean number of R. michauxii pollen grains on the body of a visitor of species X at female flowers; and fidelity is the mean proportion of pollen carried by species X that is from R. michauxii. Network Analysis To examine indirect interactions between R. michauxii males, females, and coflowering species via shared pollinators, we constructed and characterized a quantitative interaction network using the bipartite package (Dormann et al. 2008). We pooled all nine sites into a single, regional dataset in which plant species were rows (with R. michauxii males and females on separate rows), visitor species were columns, and cell values were the number of interactions. Visitors lacking pollen (total load size <2 grains) were excluded from analysis, because we were interested in interactions that had potential to transfer pollen within or between species. Within this subset of visitors, we scored each one as interacting with a given plant species if it was caught on flowers of that species and/or if it carried pollen of that species on its body. However, because flower visitors may pick up heterospecific pollen dropped on flowers by prior visitors, they can carry pollen of species they do not visit and thus to which they cannot transfer pollen (Bosch et al. 2009). To avoid inferring spurious interactions, we considered only pollen types that made up 3% or more of a visitor’s total pollen load; we also excluded singleton grains regardless of total load size. The 3% value was derived from analysis of pollen reference samples collected directly from anthers of open flowers. In 37 pollen samples from 28 plant species, each contaminating pollen species was 2.9 ± 0.9% (mean ± SE) of the total sample. Contaminating pollen on a visitor’s body would be diluted by subsequent visits to additional flowers, so we considered 3% to be a conservative threshold for evidence of a direct interaction between plant and visitor. To characterize the network, we used the bipartite package to compute several metrics that describe the role of R. michauxii within the network, or relate to whole-network stability. First, d’ describes species-level specialization—the extent to which a species links to partners nonrandomly relative to their availability in the environment (Blüthgen et al. 2006). The value of d’ ranges from 0 to 1, where low values represent generalization (random sampling of partners) and high values represent specialization. To compute d’ for R. michauxii, we assumed visitor ‘availability’ included the total number of specimens caught on any flower, regardless of the plant species or pollen load presence and composition, then asked whether actual pollen vectors interacting with R. michauxii were a nonrandom subset of this community. We examined the role of R. michauxii males and females in the network via modularity analysis. Modularity describes the extent to which a network is partitioned into modules, or groups of species that interact more with each other than with other species. We computed network-level modularity and module membership using the method of Beckett (2016), implemented in the ‘computeModules’ function. To classify the role of R. michauxii within the network, we computed its among-module connectivity (c) and within-module degree (z) and used critical values of c and z established by Olesen et al. (2007) to classify its role as a network hub, module hub, peripheral, or connector species. (Using weighted versus unweighted versions of c and z did not qualitatively alter the results; we present unweighted versions.) Whole-network properties thought to confer stability upon the network, while being reasonably robust to network size and sampling completeness, include specialization and nestedness. H2’ is a measure of specialization, the network-level analog of d’ above (Blüthgen et al. 2006). For nestedness, we used the metric ‘weighted nestedness,’ which considers not only the existence of an interaction but its frequency. Its value ranges from 0 to 1, where 1 is perfectly nested, meaning specialists interact only with a subset of the species with which generalists interact (Galeano et al. 2009). In addition, we computed Shannon diversity and Shannon evenness of interactions; larger values are thought to confer stability upon a network, but these metrics are sensitive to network size and sampling and should not typically be compared directly between networks (Kaiser-Bunbury and Blüthgen 2015). We tested the significance of network-level modularity, specialization, nestedness, interaction diversity, and interaction evenness using a null model analysis with the conservative ‘swap.web’ algorithm to generate 1,000 null networks that retained the marginal totals and connectance (number of links/species combinations) of the original network (Dormann et al. 2009). Time-lapse Video To assess differences between male and female plants in duration, rate, and diurnal timing of bee visitation, we constructed two generalized linear mixed models in the lme4 package (Bates et al. 2015), assessed model fits using the diagnostic plots of the DHARMa package, and tabulated estimated marginal means using the emmeans package. First, to compare visit duration between sexes and visitor taxa, we modeled visit duration as a function of plant sex, bee taxon, and their interaction. We modeled visit duration (measured in number of video frames) with a Poisson error distribution and log link function. The model included a random effect of site (which in most cases was equivalent to a random effect of plant, since there was one sample per site at five of the eight sites), as well as an observation-level random effect to correct overdispersion. (This model fit the data better than did a version with negative binomial error and no observation-level random effect.) Second, to create the visit frequency dataset, we tabulated visits to each plant by visitor taxon and hour of day. Zeros were scored only for taxa that were present at a site but absent during some hours. (If a taxon never occurred at a site, it simply contributed no data for that site.) We then modeled the number of visits as a function of plant sex, visitor taxon, and hour of day, as well as their two- and three-way interactions, and included a random effect of site as well as an offset variable to account for differences in the number of minutes scored per plant per hour (due to subsampling busy footage or excluding periods of rain). We modeled hour of day as an orthogonal polynomial to allow for a hump-shaped response in which there were many visits in the middle of the day but few early and late. We modeled the number of visits with negative binomial error distribution and log link function. The three-way interaction between hour, sex, and taxon was not significant (P > 0.05) and we present results of the simplified model with this term removed. Results Pollinator Community and Pollen Transport In 31.7 person-hours of observation at 10 sites, we accumulated a sample of 288 arthropods captured at R. michauxii flowers, representing 55 species or morphospecies, of which 28 were singletons (Supp Table 3 [online only]). The value of the Chao1 diversity estimator was 94 (95% CI 70–157), indicating that with continued sampling we would expect to find at least 94 species visiting R. michauxii flowers in our study area. (Because about half the detected species were singletons, Chao1 may provide a lower bound, rather than an estimate, of actual diversity.) Visitor species accumulated faster on male flowers than on female flowers, although 95% confidence intervals of the accumulation curves narrowly overlapped (Supp Fig. 1 [online only]). Our sample included seven arthropod orders, composed of 19 insect families and two spider families. The majority (82% of specimens and 45% of species) were bees (Fig. 2). At the order level, flower visitors did not differ in the likelihood of carrying pollen on their bodies (Fisher’s exact test, P = 0.09, Fig. 2), but small sample sizes in nonbee taxa (even after excluding orders with n < 10) provided low power to detect differences. Fig. 2. View largeDownload slide Composition of the R. michauxii flower-visitor community at the order level (pooled across male and female flowers). Within the order Hymenoptera, bees and wasps are shown separately. Values over bars indicate the percent of specimens in each category that had R. michauxii pollen on their bodies. Differences between orders in proportion of insects carrying pollen were not significant (exact contingency analysis, P = 0.09). Fig. 2. View largeDownload slide Composition of the R. michauxii flower-visitor community at the order level (pooled across male and female flowers). Within the order Hymenoptera, bees and wasps are shown separately. Values over bars indicate the percent of specimens in each category that had R. michauxii pollen on their bodies. Differences between orders in proportion of insects carrying pollen were not significant (exact contingency analysis, P = 0.09). The composition of pollen on flower-visitors’ bodies differed significantly between visitors collected at male flowers, female flowers at mixed sites, and female flowers at all-female sites (Fig. 3, Fisher’s exact test, P < 0.001). A lower proportion of visitors carried pollen at female flowers than at male flowers; when they did, it was more often mixed with pollen of other species (Fig. 3). Of the 11 visitors collected at all-female sites, none carried R. michauxii pollen; this was still true when we considered the additional 16 visitors collected at an all-female site in 2016 (Supp Table 1 and Supp Fig. 2 [online only]). Fig. 3. View largeDownload slide Rhus michauxii pollen was less likely to be present or pure on the bodies of visitors to female R. michauxii flowers than to male flowers; visitors to female flowers at all-female sites never carried R. michauxii pollen (see also Supp Fig. 2 [online only]). Fig. 3. View largeDownload slide Rhus michauxii pollen was less likely to be present or pure on the bodies of visitors to female R. michauxii flowers than to male flowers; visitors to female flowers at all-female sites never carried R. michauxii pollen (see also Supp Fig. 2 [online only]). Lack of pollen vectors at female flowers could arise if male and female flowers attract different visitor species. Of the 27 nonsingleton species, six were found only on male flowers and three only on female flowers; these included some of the most common species collected. For example, Augochlorella gratiosa (Smith) (Hymenoptera: Halictidae) (n = 11) and Lasioglossum reticulatum (Robertson) (Hymenoptera: Halictidae) (n = 12) were found only on male flowers, and the cuckoo bee Epeolus lectoides Robertson (Hymenoptera: Apidae) (n = 17) only on female flowers. Although A. gratiosa was found at a single, all-male site, the other two species occurred at mixed sites but were nevertheless collected from plants of a single sex. Despite this differentiation, there was extensive community overlap between visitors to male and female plants, with Pianka’s O = 0.69 (on a scale from 0 to 1, where 1 indicates perfect overlap), a value significantly larger than expected by chance given the composition of the community overall (null model analysis P < 0.001). For species that did visit female flowers, the amount of pollen on a visitor’s body depended on visitor identity (χ2 = 53.5, df = 7, P < 0.001), plant sex (χ2 = 38.6, df = 1, P < 0.001), and their nearly significant interaction (χ2 = 12.3, df = 5, P = 0.056). At male flowers, we detected no difference among visitors in the amount of pollen carried. At female flowers, the total amount of pollen was, on average, four times less, but this reduction varied with visitor identity, leading to species-specific differences in the amount of pollen available at female flowers (Fig. 4). This analysis excluded one statistical outlier, an extraordinarily large pollen sample from a honey bee caught at a male flower; the sample may have accidentally included pollen from corbiculae. Fig. 4. View largeDownload slide Mean number of R. michauxii pollen grains carried by seven species of bees when visiting male (black/ left group for each species) or female (gray/ right group for each species) R. michauxii flowers. Note that male and female bees are analyzed separately. Light gray circles show raw data; error bars are 95% confidence intervals; groups sharing a common letter designation are not significantly different. Means are not shown for Megachile texana male bees on male flowers (n = 1) or Epeolus lectoides on male flowers (n = 0). Fig. 4. View largeDownload slide Mean number of R. michauxii pollen grains carried by seven species of bees when visiting male (black/ left group for each species) or female (gray/ right group for each species) R. michauxii flowers. Note that male and female bees are analyzed separately. Light gray circles show raw data; error bars are 95% confidence intervals; groups sharing a common letter designation are not significantly different. Means are not shown for Megachile texana male bees on male flowers (n = 1) or Epeolus lectoides on male flowers (n = 0). Finally, we asked which flower-visitor species are most likely to be important pollinators by combining information about the relative abundance, pollen load size, and pollen load purity of visitors to female plants. These aspects of PI are summarized in Table 1 and synthesized in the PI index, which suggests that M. texana females and M. mendica males were the most important vectors of R. michauxii pollen at our study sites in 2017. Table 1. Aspects of pollinator importance for the eight most abundant potential pollinator taxa caught at female flowers (note that male and female bees of the same species are treated separately) Taxon No. caught at female flowers Proportion with R. michauxii pollen Relative abundance Mean proportion R. michauxii pollen ± SE No. grainsR. michauxii pollen PI index Megachile texana F 19 0.74 0.18 0.65 ± 0.10 410 48.20 Megachile mendica M 24 0.63 0.19 0.85 ± 0.05 76 12.58 Megachile petulans M 4 0.67 0.04 0.82 ± 0.11 184 5.87 Apis mellifera W 8 0.50 0.05 0.77 ± 0.14 102 4.06 Megachile texana M 9 0.78 0.09 0.60 ± 0.14 61 3.30 Augochlorella aurata F 19 0.21 0.05 0.82 ± 0.07 38 1.63 Colletes nudus F 11 0.64 0.09 0.29 ± 0.13 56 1.46 Epeolus lectoides 17 0.24 0.05 0.47 ± 0.24 1 0.04 Taxon No. caught at female flowers Proportion with R. michauxii pollen Relative abundance Mean proportion R. michauxii pollen ± SE No. grainsR. michauxii pollen PI index Megachile texana F 19 0.74 0.18 0.65 ± 0.10 410 48.20 Megachile mendica M 24 0.63 0.19 0.85 ± 0.05 76 12.58 Megachile petulans M 4 0.67 0.04 0.82 ± 0.11 184 5.87 Apis mellifera W 8 0.50 0.05 0.77 ± 0.14 102 4.06 Megachile texana M 9 0.78 0.09 0.60 ± 0.14 61 3.30 Augochlorella aurata F 19 0.21 0.05 0.82 ± 0.07 38 1.63 Colletes nudus F 11 0.64 0.09 0.29 ± 0.13 56 1.46 Epeolus lectoides 17 0.24 0.05 0.47 ± 0.24 1 0.04 PI index summarizes pollinator importance, with larger values representing greater importance. View Large Table 1. Aspects of pollinator importance for the eight most abundant potential pollinator taxa caught at female flowers (note that male and female bees of the same species are treated separately) Taxon No. caught at female flowers Proportion with R. michauxii pollen Relative abundance Mean proportion R. michauxii pollen ± SE No. grainsR. michauxii pollen PI index Megachile texana F 19 0.74 0.18 0.65 ± 0.10 410 48.20 Megachile mendica M 24 0.63 0.19 0.85 ± 0.05 76 12.58 Megachile petulans M 4 0.67 0.04 0.82 ± 0.11 184 5.87 Apis mellifera W 8 0.50 0.05 0.77 ± 0.14 102 4.06 Megachile texana M 9 0.78 0.09 0.60 ± 0.14 61 3.30 Augochlorella aurata F 19 0.21 0.05 0.82 ± 0.07 38 1.63 Colletes nudus F 11 0.64 0.09 0.29 ± 0.13 56 1.46 Epeolus lectoides 17 0.24 0.05 0.47 ± 0.24 1 0.04 Taxon No. caught at female flowers Proportion with R. michauxii pollen Relative abundance Mean proportion R. michauxii pollen ± SE No. grainsR. michauxii pollen PI index Megachile texana F 19 0.74 0.18 0.65 ± 0.10 410 48.20 Megachile mendica M 24 0.63 0.19 0.85 ± 0.05 76 12.58 Megachile petulans M 4 0.67 0.04 0.82 ± 0.11 184 5.87 Apis mellifera W 8 0.50 0.05 0.77 ± 0.14 102 4.06 Megachile texana M 9 0.78 0.09 0.60 ± 0.14 61 3.30 Augochlorella aurata F 19 0.21 0.05 0.82 ± 0.07 38 1.63 Colletes nudus F 11 0.64 0.09 0.29 ± 0.13 56 1.46 Epeolus lectoides 17 0.24 0.05 0.47 ± 0.24 1 0.04 PI index summarizes pollinator importance, with larger values representing greater importance. View Large Network During 18 h of sampling at nine sites (see Supp Table 1 [online only] for site characteristics), we captured 313 arthropods of 94 species visiting flowers of 22 kinds (21 species, with R. michauxii male and female considered separately). Overall, R. michauxii was visited by 120 specimens (38% of the total) of 32 species (34% of the total). A subset of 256 specimens of 73 species bore pollen loads of at least two grains and were included in the interaction network (Fig. 5). Most individual pollen loads evidenced visits to one or two plant species, occasionally up to five (Supp Table 4 [online only]). Combined evidence from pollen and observed visits (by pollen-bearing insects) yielded a total of 424 interactions with 46 pollen species/morphospecies via 206 unique links (Fig. 5). Within this network, R. michauxii males interacted with 84 specimens—80 of which carried its pollen—and 21 species, representing 10% of the network links. Indeed, male R. michauxii was one of the most generalized plants in the network, ranking third of the 46 plant types in normalized degree (a measure of partner diversity). Females interacted with 28 specimens and were slightly more specialized than males. The value of d’, which represents specialization on a scale from 0 to 1 where 0 is most generalized, was 0.13 for males and 0.22 for females. Fig. 5. View largeDownload slide Quantitative interaction network for R. michauxii. Bars on the left represent plant species, with bar heights proportional to the number of flowers counted on transects (flowers not observed on transects are assigned an abundance one-half that of the least abundant on-transect species). Bars on the right represent visitor species, with bar heights proportional to the number of specimens that had pollen on their bodies. Visitor color indicates taxonomic order; links to R. michauxii males are black and to R. michauxii females, brick red. For clarity, only selected species are labeled, emphasizing those that are common or well connected. Fig. 5. View largeDownload slide Quantitative interaction network for R. michauxii. Bars on the left represent plant species, with bar heights proportional to the number of flowers counted on transects (flowers not observed on transects are assigned an abundance one-half that of the least abundant on-transect species). Bars on the right represent visitor species, with bar heights proportional to the number of specimens that had pollen on their bodies. Visitor color indicates taxonomic order; links to R. michauxii males are black and to R. michauxii females, brick red. For clarity, only selected species are labeled, emphasizing those that are common or well connected. The network was more modular than expected under the null model (Table 2). R. michauxii males and females were members of the same module, in which males were a module hub and females were peripheral. The module also included 10 other plant species and 15 visitor species (Supp Table 5 [online only]). Network-level specialization (H’) was greater than expected by chance, nestedness did not differ from the null expectation, and interaction diversity and evenness were both lower than expected by chance (Table 2). Table 2. Network metrics and their statistical significance for the interaction network containing R. michauxii Metric Observed value Null mean P Departure from null Number of modules 6 na na Modularity 0.53 0.44 <0.001 higher Specialization (H2’) 0.38 0.34 0.001 higher Weighted nestedness 0.66 0.62 0.190 NS Shannon diversity of interactions 4.96 5.04 0.001 lower Interaction evenness 0.61 0.62 0.001 lower Metric Observed value Null mean P Departure from null Number of modules 6 na na Modularity 0.53 0.44 <0.001 higher Specialization (H2’) 0.38 0.34 0.001 higher Weighted nestedness 0.66 0.62 0.190 NS Shannon diversity of interactions 4.96 5.04 0.001 lower Interaction evenness 0.61 0.62 0.001 lower View Large Table 2. Network metrics and their statistical significance for the interaction network containing R. michauxii Metric Observed value Null mean P Departure from null Number of modules 6 na na Modularity 0.53 0.44 <0.001 higher Specialization (H2’) 0.38 0.34 0.001 higher Weighted nestedness 0.66 0.62 0.190 NS Shannon diversity of interactions 4.96 5.04 0.001 lower Interaction evenness 0.61 0.62 0.001 lower Metric Observed value Null mean P Departure from null Number of modules 6 na na Modularity 0.53 0.44 <0.001 higher Specialization (H2’) 0.38 0.34 0.001 higher Weighted nestedness 0.66 0.62 0.190 NS Shannon diversity of interactions 4.96 5.04 0.001 lower Interaction evenness 0.61 0.62 0.001 lower View Large Time-lapse Video We scored time-lapse video representing 313 daylight hours, including 2,596 visits to 11 plants (Supp Video 1 [online only]). (The two pooled males from a single site are here treated as one sample or ‘plant.’) Of the total visits, 88% (2,293 visits) were by bees. Of the visits by bees, 99% (2,259) were identified at the genus or species level and, for Megachile, also sexed. Some taxa appeared in a single visit to a single plant (Halictus, Bombus) or on male plants only (Lasioglossum vierecki [Crawford]). These taxa were excluded from analysis, such that 2,213 visits by seven taxa (Apis mellifera L., Colletes nudus Robertson, Augochlorella, Augochloropsis, dark Lasioglossum, Megachile female, and Megachile male) were included in the comparison of visit duration, frequency, and diurnal timing on male and female plants. Bee visit duration depended on visitor taxon (χ2 = 275, df = 6, P < 0.001; Supp Fig. 3 [online only]), but not plant sex (χ2 = 0.296, df = 1, P = 0.59) or sex–taxon interaction (χ2 = 5.11, df = 6, P = 0.53). Megachile males and females made the shortest visits (estimated marginal mean 9s for males, 14s for females) and halictids made the longest, often walking around on the inflorescence for minutes at a time (estimated marginal mean 68s for Lasioglossum, 39s for Augochlorella, 29s for Augochloropsis). Visits by Apis and Colletes were intermediate in duration (Supp Fig. 3 [online only]). Bee visit frequency depended on plant sex, visitor taxon, and time of day, as well as sex–taxon and taxon–hour interactions (Table 3). Visitation to both male and female flowers peaked during mid-day, with model estimates reaching maxima at 12:00–13:00 for males and at 13:00–14:00 for females. There was no sex–hour interaction, indicating that the diurnal pattern of visits was similar for male and female inflorescences. Females were, however, visited less often than males (Fig. 6). Averaged over all taxa and all hours from 7:00 to 19:00 inclusive, females received six times fewer visits than males, with 0.2 (95% CI 0.2–0.4) visits per taxon per hour for females compared 1.5 (1.2–1.9) for males. In addition, the sex–taxon interaction indicates that the difference in visitation frequency to males and females was not consistent across visitor taxa. The preference for males was most pronounced in Augochlorella, but followed a similar trend in all taxa except Lasioglossum, which visited females slightly (and nonsignficantly) more often than males (Supp Fig. 4 [online only]). Table 3. Results of Wald χ2 tests for effect of plant sex, visitor taxon, and time of day on visitation rate to R. michauxii inflorescences χ2 df P Sex 33.6 1 <0.001 Taxon 35.8 6 <0.001 Houra 49.6 2 <0.001 Sex:Taxon 37.5 6 <0.001 Sex:Hour 4.54 2 0.103 Taxon:Hour 24.0 12 0.020 χ2 df P Sex 33.6 1 <0.001 Taxon 35.8 6 <0.001 Houra 49.6 2 <0.001 Sex:Taxon 37.5 6 <0.001 Sex:Hour 4.54 2 0.103 Taxon:Hour 24.0 12 0.020 aHour represents an orthogonal polynomial term that allows for a hump-shaped relationship between visitation rate and time of day. View Large Table 3. Results of Wald χ2 tests for effect of plant sex, visitor taxon, and time of day on visitation rate to R. michauxii inflorescences χ2 df P Sex 33.6 1 <0.001 Taxon 35.8 6 <0.001 Houra 49.6 2 <0.001 Sex:Taxon 37.5 6 <0.001 Sex:Hour 4.54 2 0.103 Taxon:Hour 24.0 12 0.020 χ2 df P Sex 33.6 1 <0.001 Taxon 35.8 6 <0.001 Houra 49.6 2 <0.001 Sex:Taxon 37.5 6 <0.001 Sex:Hour 4.54 2 0.103 Taxon:Hour 24.0 12 0.020 aHour represents an orthogonal polynomial term that allows for a hump-shaped relationship between visitation rate and time of day. View Large Fig. 6. View largeDownload slide Visitation by seven common bee taxa to R. michauxii male (a) and female (b) inflorescences in time-lapse video; values for each hour are averaged across plants (n = 7 males, 4 females). Females were visited significantly less often (c); values are model-estimated means for each sex/hour combination, averaged across visitor taxa, with 95% confidence intervals. Overlapping values are jittered for clarity. Fig. 6. View largeDownload slide Visitation by seven common bee taxa to R. michauxii male (a) and female (b) inflorescences in time-lapse video; values for each hour are averaged across plants (n = 7 males, 4 females). Females were visited significantly less often (c); values are model-estimated means for each sex/hour combination, averaged across visitor taxa, with 95% confidence intervals. Overlapping values are jittered for clarity. Discussion Rarity and dioecy pose challenges to sexual reproduction in plants by reducing the abundance or effectiveness of pollinator visits, particularly to female flowers. We show that, despite its regional rarity, the federally endangered, dioecious shrub R. michauxii attracted a large proportion of available flower visitors in its local environment, and these visitors collected large pollen loads from male flowers. As expected, female inflorescences were visited less often than males, but broad overlap between the visitor communities of the two sexes suggests that pollinator visitation per se is not a primary barrier to sexual reproduction in this species. Instead, we demonstrate greatly reduced and mixed pollen loads on visitors to female flowers, and an absence of conspecific pollen at all-female sites. Preliminary results (to be presented elsewhere) further indicate that plants in all-female sites neither received conspecific pollen nor developed mature seeds, corroborating the pattern of pollen transport by insects reported here. Although reproductive failure at all-male sites is less obvious, it is likely that their pollen is largely or entirely wasted due to a lack of appropriate recipients. R. michauxii attracted 38% of all individual flower visitors sampled from the study sites, and R. michauxii pollen was included in 31% of individual pollen loads, suggesting that it competes successfully with coflowering plants to attract flower visitors. Indeed, R. michauxii flowers were among the most abundant flowers in our study, ranking first or second in abundance at each site, and second in the regional sample pooled across study sites (Fig. 5). Thus, unlike some rare plants whose local abundance is too low to support pollinator populations—which must, therefore, subsist primarily on coflowering species (e.g., Sipes and Tepedino 1995, Gibson et al. 2006, Larson et al. 2014)—R. michauxii itself appears to provide important resources for insects. Nevertheless, its bloom period is short—less than a month at any one site—indicating that other floral resources throughout the season are essential to support the pollinator community. Future studies should identify the species that facilitate R. michauxii by provisioning its pollinators outside its flowering period. Analysis of visitor abundance and pollen load size and purity indicated that bees in the genus Megachile (Hymenoptera: Megachilidae) were the flower visitors most likely to bring conspecific pollen to female plants (Table 1). In addition, nine of 11 plants subject to time-lapse videography received visits from Megachile bees during the 2 d of footage analyzed, confirming their consistent presence across sites and sexes. The abundance of megachilids in our sample suggests that their habitat requirements are currently being met at the study sites. Nevertheless, attention to their foraging and nesting resources may promote success of future habitat management or restoration efforts for R. michauxii. The three most important potential pollinator species (Megachile texana Cresson, Megachile mendica Cresson, and Megachilepetulans Cresson) are broad dietary generalists (Mitchell 1960) that require floral resources before and after peak R. michauxii flowering in June. In a 2-yr survey of bees in nearby sites in 2012 and 2013, M. mendica was found from mid-April to mid-October, M. texana from mid-May to mid-September, and M. petulans from late May to mid-August (Moylett 2014). Their nesting resources vary across species. M. texana nests in the ground, and may excavate nests or use pre-existing insect burrows, on the soil surface under rocks or in tunnels 4–5 cm deep (Krombein 1970, Eickwort et al. 1981). M. mendica may nest in soil or in hollow stems such as those of rose and sumac (Krombein 1967, Williams et al. 1986), while M. petulans uses pre-existing cavities in wood or stems (Graham et al. 2014). Although megachilid bees were important potential pollinators of R. michauxii, the flower-visitor community was overall highly diverse—our sample included 55 species or morphospecies of arthropods, including 26 bee species—and R. michauxii was among the most generalized plant species in the interaction network. Many flower-visitor species were capable of transporting R. michauxii pollen, with more than 80% of visitors to male inflorescences acquiring pollen loads prior to being captured. Given interspecific differences in foraging behavior—such as differences in visit duration (Supp Fig. 3 [online only]) or patterns of movement across inflorescences—diverse visitors may complement one another to provide overall more effective pollination than would any one species alone (e.g., Chagnon et al. 1993). The results of R. michauxii targeted sampling, network analysis, and time-lapse videography all support the expectation that female flowers would be less attractive than males to insect visitors (e.g., Bierzychudek 1987, Kevan et al. 1990, but see Kay et al. 1984, Ashman 2000). Two of the most abundant visitor species—Augochlorella gratiosa and Lasioglossum reticulatum—were found exclusively on male plants, and species accumulation curves showed that visitor richness was marginally higher on male plants (Supp Fig. 1 [online only]). Time-lapse video indicated that, averaged over taxa and hours, bees visited male inflorescences about six times more often than female inflorescences. Despite these differences, there was broad overlap between the insect communities visiting male and female flowers, and both sexes were members of the same network module—where a module is a group of species that interact more with one another than with other taxa in the network. Within this module, however, males took on the role of module hub, indicating extensive interaction with most other module members, whereas females had a peripheral role, indicating less extensive connection within their module and few, if any, outside it (Olesen et al. 2007). These patterns suggest that, consistent with the prediction for female flowers more generally, female R. michauxii flowers could be susceptible to insufficient visitation if the pollinator community were to decline. Counter to expectations, however, male and female inflorescences did not differ in the duration or diurnal timing of bee visits. In many dioecious systems, visits to female flowers are brief, particularly in cases where female flowers offer little or no reward (e.g., Ågren et al. 1986, Bierzychudek 1987). In these cases, generalist pollinators are also expected to visit male flowers first, and females later in the day (Bawa 1980, Beach 1981). Detailed assessment of the nectar and pollen resources provided by R. michauxii remains for future studies, but insect behavior and the large, visible nectaries of female flowers suggest that they do provide substantial resources—a feature that may be typical of the genus Rhus (Young 1972, Greco et al. 1996, Matsuyama et al. 2009). Other Rhus species, however, appear to vary in the extent of pollinator preference for male flowers. Young (1972) suggested, anecdotally, that honey bees did not distinguish between hermaphrodite and female flowers of Rhus integrifolia and R. ovata in California. Matsuyama et al. (2009) documented male preference in many, but not all, of the diverse visitors to Rhus trichocarpa in Japan. And Greco et al. (1996) showed that male Rhus hirta flowers provide pollen in the morning but secrete little nectar, whereas female flowers of the same species secrete nectar in the afternoon. Honey bees, the sole visitor in that study, were thus attracted to males in the morning, but occasionally dropped pollen on female plants during pollen-foraging bouts. Then nectar foragers visited female flowers in the afternoon and further distributed previously dropped pollen among the stigmas. Further comparative work could suggest whether visitation patterns observed in R. michauxii contribute to any differences in its pollination success relative to its more widespread congeners. Taken together, our results suggest that existing plant–pollinator interactions could support sexual reproduction in R. michauxii, but that they fail to do so in single-sex sites. Restoration of widespread sexual reproduction of this species in the sandhills region will require facilitated mating, such as hand pollination (e.g., Ye et al. 2007, Kawelo et al. 2012) or the identification and translocation of suitable mates into single-sex sites (Orsenigo et al. 2017). Before such efforts are undertaken, more research is needed to assess their likely success. Outbreeding depression and disruption of locally adapted ecotypes are concerns when translocating plants for conservation or restoration, particularly for species with strong spatial genetic structure (Hufford and Mazer 2003). An analysis of R. michauxii populations in North Carolina demonstrated that a relatively high proportion (34%) of genetic variation was partitioned between, rather than within, sites (Sherman-Broyles et al. 1992), indicating that potential transplants should be chosen carefully on the basis of similarity in genotype and local habitat, as well as experimental crosses and their outcomes for seed set and seedling viability. The case of R. michauxii highlights conservation challenges produced by the dioecious mating system in sessile organisms. Although dioecious species are only a small percentage of the world’s plant diversity, they are overrepresented on lists of rare or threatened taxa and are likely to suffer increasingly from effects of anthropogenic habitat fragmentation. Even when the resulting small, isolated populations succeed in attracting pollinators, further intervention may be needed to ensure sexual reproduction and evolutionary potential in these species. Acknowledgments John McAllister, Janet Gray, Brady Beck, and Fort Bragg Range Control facilitated site access. Madeline Adams, Riki Bonnema, Francesca Romero, and Ryan Warren collected insects and processed samples and video. Matt Bertone and Sam Droege assisted with insect identification. Dale Suiter, Jenny Stanley, and Lesley Starke assisted with federal and state research permits for endangered plants. This work was funded by the U.S. Department of Defense and NC State University. References Cited Ågren J. 1996 . Population size, pollinator limitation, and seed set in the self‐incompatible herb Lythrum salicaria . Ecology 77 : 1779 – 1790 . Google Scholar Crossref Search ADS Ågren J. , T. Elmqvist , and A. Tunlid . 1986 . Pollination by deceit, floral sex ratios and seed set in dioecious Rubus chamaemorus L . Oecologia 70 : 332 – 338 . Google Scholar Crossref Search ADS PubMed Aguilar R. , L. Ashworth , L. Galetto , and M. A. Aizen . 2006 . Plant reproductive susceptibility to habitat fragmentation: review and synthesis through a meta-analysis . Ecol. Lett . 9 : 968 – 980 . Google Scholar Crossref Search ADS PubMed Arnett R. H. , M. C. Thomas , P. E. Skelley , and J. H. Frank . 2002 . American beetles, vol. II: polyphaga: scarabaeoidea through curculionoidea , CRC Press , Boca Raton, FL . Ascher J. S. , and J. Pickering . 2014 . Discover life bee species guide and world checklist (Hymenoptera: Apoidea: Anthophila) (Draft 39, 22 April 2014) . http://www.discoverlife.org/mp/20q?guide=Apoidea_species Ashman T.-L. 2000 . Pollinator selectivity and its implications for the evolution of dioecy and sexual dimorphism . Ecology 81 : 2577 – 2591 . Google Scholar Crossref Search ADS Bates D. , M. Maechler , B. Bolker , and S. Walker . 2015 . Fitting linear mixed-effects models using lme4 . J. Stat. Softw . 67 : 1 – 48 . Google Scholar Crossref Search ADS Bawa K. S. 1980 . Evolution of dioecy in flowering plants . Annu. Rev. Ecol. Syst . 11 : 15 – 39 . Google Scholar Crossref Search ADS Beach J. H. 1981 . Pollinator foraging and the evolution of dioecy . Am. Nat . 118 : 572 – 577 . Google Scholar Crossref Search ADS Beckett S. J. 2016 . Improved community detection in weighted bipartite networks . R. Soc. Open Sci . 3 : 140536 . Google Scholar Crossref Search ADS PubMed Bierzychudek P. 1987 . Pollinators increase the cost of sex by avoiding female flowers . Ecology 68 : 444 – 447 . Google Scholar Crossref Search ADS Blüthgen N. , F. Menzel , and N. Blüthgen . 2006 . Measuring specialization in species interaction networks . BMC Ecol . 6 : 9 . Google Scholar Crossref Search ADS PubMed Bohart R. 1982 . A synopsis of the Chrysididae in America north of Mexico . Mem. Am. Entomol. Inst . 33 : 1 – 266 . Bohart R. , and L. Kimsey . 1980 . A generic synopsis of the Chrysididae of America North of Mexico (Hymenoptera) . J. Kans. Entomol. Soc . 137 – 148 . Bohart R. M. , and A. S. Menke . 1976 . Sphecid wasps of the world: a generic revision , University of California Press , Berkeley, CA . Bosch J. , A. M. González , A. Rodrigo , and D. Navarro . 2009 . Plant–pollinator networks: adding the pollinator’s perspective . Ecol. Lett . 12 : 409 – 419 . Google Scholar Crossref Search ADS PubMed Bradley J. C. 1957 . The taxa of Campsomeris (Hymenoptera: Scoliidae) occurring in the New World . Trans. Am. Entomol. Soc . 83 : 65 – 77 . Braham R. , C. Murray , and M. Boyer . 2006 . Mitigating impacts to Michaux’s Sumac (Rhus michauxii Sarg.): a case study of transplanting an endangered shrub . Castanea 71 : 265 – 271 . Google Scholar Crossref Search ADS Brown P. J. , K. R. Wormington , and P. Brown . 2015 . Identifying essential ecological factors underpinning the development of a conservation plan for the endangered Australian tree Alectryon ramiflorus . Oryx 49 : 453 – 460 . Google Scholar Crossref Search ADS Carlsson-Graner U. , T. Elmqvist , J. Agren , H. Gardfjell , and P. Ingvarsson . 1998 . Floral sex ratios, disease and seed set in dioecious Silene dioica . J. Ecol . 86 : 79 – 91 . Google Scholar Crossref Search ADS Chagnon M. , J. Gingras , and D. DeOliveira . 1993 . Complementary aspects of strawberry pollination by honey and indigenous bees (Hymenoptera) . J. Econ. Entomol . 86 : 416 – 420 . Google Scholar Crossref Search ADS Charlesworth D. 1993 . Why are unisexual flowers associated with wind pollination and unspecialized pollinators? Am. Nat . 141 : 481 – 490 . Google Scholar Crossref Search ADS Dormann C. F. , B. Gruber , and J. Fruend . 2008 . Introducing the bipartite package: analysing ecological networks . R News . 8 : 8 – 11 . Dormann C. F. , J. Fründ , N. Blüthgen , and B. Gruber . 2009 . Indices, graphs and null models: analyzing bipartite ecological networks . Open Ecol. J . 2 : 7 – 24 . Google Scholar Crossref Search ADS Edwards J. , G. P. Smith , and M. H. McEntee . 2015 . Long-term time-lapse video provides near complete records of floral visitation . J. Pollination Ecol . 16 : 91 – 100 . Eickwort G. C. , R. W. Matthews , and J. Carpenter . 1981 . Observations on the nesting behavior of Megachile rubi and M. texana with a discussion of the significance of soil nesting in the evolution of megachilid bees (Hymenoptera: Megachilidae) . J. Kans. Entomol. Soc . 54 : 557 – 570 . Enns W. R. 1955 . A revision of the genera Nemognatha, Zonitis, and Pseudozonitis (Coleoptera, Meloidae) in America north of Mexico, with a proposed new genus . Univ. Kansas Sci. Bull . 37 : 685 – 909 . Fall H. C. 1899 . Synopsis of the species of Acmaeodera of America north of Mexico . J. N. Y. Entomol. Soc . 7 : 1 – 37 . Galeano J. , J. M. Pastor , and J. M. Iriondo . 2009 . Weighted-interaction nestedness estimator (WINE): a new estimator to calculate over frequency matrices . Environ. Model. Software 24 : 1342 – 1346 . Google Scholar Crossref Search ADS Gibbs J. 2011 . Revision of the metallic Lasioglossum (Dialictus) of eastern North America (Hymenoptera: Halictidae: Halictini) . Zootaxa 3073 : 1 – 216 . Google Scholar Crossref Search ADS Gibson R. , I. Nelson , G. Hopkins , B. Hamlett , and J. Memmott . 2006 . Pollinator webs, plant communities and the conservation of rare plants: arable weeds as a case study . J. Appl. Ecol . 43 : 246 – 257 . Google Scholar Crossref Search ADS Gotelli N. J. , E. M. Hart , and A. M. Ellison . 2015 . EcoSimR: Null model analysis for ecological data. R package version 0.1.0 . doi: https://doi.org/10.5281/zenodo.16522 . Graham J. R. , Q. Tan , L. C. Jones , and J. D. Ellis . 2014 . Native Buzz: citizen scientists creating nesting habitat for solitary bees and wasps . Fla. Sci . 74 : 204 – 218 . Greco C. F. , D. Holland , and P. G. Kevan . 1996 . Foraging behaviour of honey bees (Apis mellifera L.) on staghorn sumac [Rhus hirta Sudworth (ex-typhina L.)]: differences and dioecy . Can. Entomol . 128 : 355 – 366 . Google Scholar Crossref Search ADS Hall J. C. , and N. L. Evenhuis . 2003 . Review of the subgenus Geron (Geron) Meigen in the Nearctic Region (Diptera: Bombyliidae: Toxophorinae) . Zootaxa 181 : 1 – 72 . Google Scholar Crossref Search ADS Hartig F. 2018 . DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models R package version 0.1.6 . http://florianhartig.github.io/DHARMa/ Heilbuth J. C. 2000 . Lower species richness in dioecious clades . Am. Nat . 156 : 221 – 241 . Google Scholar Crossref Search ADS PubMed House S. M. 1992 . Population density and fruit set in three dioecious tree species in Australian tropical rain forest . J. Ecol . 80 : 57 – 69 . Google Scholar Crossref Search ADS House S. M. 1993 . Pollination success in a population of dioecious rain forest trees . Oecologia 96 : 555 – 561 . Google Scholar Crossref Search ADS PubMed Howden H. F. 1968 . A review of the Trichiinae of North and Central America (Coleoptera: Scarabaeidae) . Mem. Entomol. Soc. Can . 100 : 5 – 77 . Google Scholar Crossref Search ADS Hsieh T. , K. Ma , and A. Chao . 2016 . iNEXT: an R package for rarefaction and extrapolation of species diversity (Hill numbers) . Methods Ecol. Evol . 7 : 1451 – 1456 . Google Scholar Crossref Search ADS Hufford K. M. , and S. J. Mazer . 2003 . Plant ecotypes: genetic differentiation in the age of ecological restoration . Trends Ecol. Evol . 18 : 147 – 155 . Google Scholar Crossref Search ADS Kaiser-Bunbury C. N. , and N. Blüthgen . 2015 . Integrating network ecology with applied conservation: a synthesis and guide to implementation . AoB Plants . 7 : plv076 . Google Scholar Crossref Search ADS PubMed Kawelo H. K. , S. C. Harbin , S. M. Joe , M. J. Keir , and L. Weisenberger . 2012 . Unique reintroduction considerations in Hawaii: case studies from a decade of rare plant restoration at the Oahu Army Natural Resource Rare Plant Program, pp. 209 – 226 . In H. K. E. Maschinski, J. and Raven P. H. (eds.), Plant reintroduction in a changing climate: the science and practice of ecological restoration . Island Press , Washington, DC . Kay Q. , A. Lack , F. Bamber , and C. Davies . 1984 . Differences between sexes in floral morphology, nectar production and insect visits in a dioecious species, Silene dioica . New Phytol . 98 : 515 – 529 . Google Scholar Crossref Search ADS Kearns C. A. , and D. W. Inouye . 1993 . Techniques for pollination biologists , The University Press of Colorado , Niwot, CO . Kevan P. G. , D. Eisikowitch , J. D. Ambrose , and J. R. Kemp . 1990 . Cryptic dioecy and insect pollination in Rosa setigera Michx. (Rosaceae), a rare plant of Carolinian Canada . Biol. J. Linn. Soc . 40 : 229 – 243 . Google Scholar Crossref Search ADS Kits J. H. , S. A. Marshall , and N. L. Evenhuis . 2008 . The bee flies (Diptera: Bombyliidae) of Ontario, with a key to the species of eastern Canada . Can. J. Arthropod Ident . 6 : 1 – 52 . Krombein K. V. 1967 . Trap-nesting wasps and bees: life histories, nests, and associates , Smithsonian Press , Washington, DC . Krombein K. V. 1970 . Another note on the nesting habits of Megachile texana Cresson . Proc. Entomol. Soc. Wash . 72 : 415 . Kunin W. E. 1997a . Population biology and rarity: on the complexity of density dependence in insect–plant interactions, pp. 150 – 173 . In W. Kunin and K. J. Gaston (eds.), The biology of rarity: causes and consequences of rare-common differences . Chapman and Hall , New York . Kunin W. E. 1997b . Population size and density effects in pollination: pollinator foraging and plant reproductive success in experimental arrays of Brassica kaber . J. Ecol . 85 : 225 – 234 . Google Scholar Crossref Search ADS Larson D. L. , S. Droege , P. A. Rabie , J. L. Larson , J. Devalez , M. Haar , and M. McDermott‐Kubeczko . 2014 . Using a network modularity analysis to inform management of a rare endemic plant in the northern Great Plains, USA . J. Appl. Ecol . 51 : 1024 – 1032 . Google Scholar Crossref Search ADS Lenth R. , J. Love , and M. Herve . 2018 . emmeans: Estimated Marginal Means, aka Least-Squares Means R Package version 1.1.3 . https://github.com/rvlenth/emmeans Liljeblad E. 1945 . Monograph of the family Mordellidae (Coleoptera) of North America, north of Mexico, pp . 1 – 229 . Miscellaneous Publications no. 62, Museum of Zoology, University of Michigan , Ann Arbor . Linsley E. G. , and J. A. Chemsak . 1961 . The cerambycidae of North America part VI no. 2: taxonomy and classification of the subfamily lepturinae , University of California Press , Berkeley, CA . Mack A. L. 1997 . Spatial distribution, fruit production and seed removal of a rare, dioecious canopy tree species (Aglaia aff. flavida Merr. et Perr.) in Papua New Guinea . J. Trop. Ecol . 13 : 305 – 316 . Google Scholar Crossref Search ADS Matsuyama S. , N. Osawa , and M. Sakimoto . 2009 . Generalist pollinators in the dioecious shrub Rhus trichocarpa Miq. (Anacardiaceae) and their role in reproductive success . Plant Species Biol . 24 : 215 – 224 . Google Scholar Crossref Search ADS McAlpine , J. F. (ed.) 1981 . Manual of Nearctic Diptera, vol. 2 . Agriculture Canada , Ottawa, Ontario . Metcalfe D. , and W. Kunin . 2006 . The effects of plant density upon pollination success, reproductive effort and fruit parasitism in Cistus ladanifer L. (Cistaceae) . Plant Ecol . 185 : 41 . Google Scholar Crossref Search ADS Miranda G. , A. Young , M. Locke , S. Marshall , J. Skevington , and F. Thompson . 2013 . Key to the genera of Nearctic Syrphidae . Can. J. Arthropod Ident . 23 : 351 . Mitchell T. B. 1960 . Bees of the Eastern United States , North Carolina Agricultural Experiment Station , Raleigh, NC . Moylett H. M. C. 2014 . The impact of prescribed burning on native bee communities (Hymenoptera: Apoidea: Anthophila) in Longleaf Pine (Pinus palustris Mill.) Savannas in the North Carolina Sandhills . M.S. thesis, North Carolina State University . Olesen J. M. , J. Bascompte , Y. L. Dupont , and P. Jordano . 2007 . The modularity of pollination networks . Proc. Natl Acad. Sci. USA 104 : 19891 – 19896 . Google Scholar Crossref Search ADS Orsenigo S. , R. Gentili , A. J. Smolders , A. Efremov , G. Rossi , N. M. Ardenghi , S. Citterio , and T. Abeli . 2017 . Reintroduction of a dioecious aquatic macrophyte (Stratiotes aloides L.) regionally extinct in the wild. Interesting answers from genetics . Aquat. Conserv.: March Freshwat. Ecosyst . 27 : 10 – 23 . Google Scholar Crossref Search ADS Painter R. H. 1962 . The taxonomy and biology of Systoechus and Anastoechus bombyliid (Diptera) predators in grasshopper egg pods . J. Kans. Entomol. Soc . 35 : 255 – 269 . Pannell J. R. , and S. C. H. Barrett . 1998 . Baker’s law revisited: reproductive assurance in a metapopulation . Evolution 52 : 657 – 668 . Google Scholar Crossref Search ADS PubMed Pianka E. R. 1973 . The structure of lizard communities . Annu. Rev. Ecol. Syst . 4 : 53 – 74 . Google Scholar Crossref Search ADS Primack R. B. , and J. A. Silander . 1975 . Measuring the relative importance of different pollinators to plants . Nature 255 : 143 – 144 . Google Scholar Crossref Search ADS Rathcke B. 1983 . Competition and facilitation among plants for pollination, pp. 305 – 329 . In L. Real (ed.), Pollination biology . Academic Press , Orlando, FL . R Core Team . 2013 . R: A language and environment for statistical computing , R Foundation for Statistical Computing , Vienna, Austria . Renner S. S. , and R. E. Ricklefs . 1995 . Dioecy and its correlates in the flowering plants . Am. J. Bot . 82 : 596 – 606 . Google Scholar Crossref Search ADS RStudio Team . 2015 . RStudio: Integrated Development for R , RStudio, Inc. , Boston, MA. http://www.rstudio.com/ Schemske D. W. , and C. C. Horvitz . 1984 . Variation among floral visitors in pollination ability: a precondition for mutualism specialization . Science 225 : 519 – 521 . Google Scholar Crossref Search ADS PubMed Sherman-Broyles S. L. , J. P. Gibson , J. Hamrick , M. A. Bucher , and M. J. Gibson . 1992 . Comparisons of allozyme diversity among rare and widespread Rhus species . Syst. Bot . 17 : 551 – 559 . Google Scholar Crossref Search ADS Sipes S. D. , and V. J. Tepedino . 1995 . Reproductive biology of the rare orchid, Spiranthes diluvialis: breeding system, pollination, and implications for conservation . Conserv. Biol . 9 : 929 – 938 . Google Scholar Crossref Search ADS Somanathan H. , and R. M. Borges . 2000 . Influence of exploitation on population structure, spatial distribution and reproductive success of dioecious species in a fragmented cloud forest in India . Biol. Conserv . 94 : 243 – 256 . Google Scholar Crossref Search ADS Sorrie B. A. 2011 . A field guide to wildflowers of the Sandhills Region: North Carolina, South Carolina, and Georgia , Univ of North Carolina Press , Chapel Hill, NC . Stehr F. W. 1987 . Immature Insects , Kendall/Hunt , Dubuque, IA . Thompson F. C. , F. D. Fee , and L. Bezark . 1990 . Two immigrant synanthropic flower flies (Diptera: Syrphidae) new to North America . Entomol. News . 101 : 69 – 74 . Ubick D. , P. Paquin , P. E. Cushing , and V. Roth (eds.). 2005 . Spiders of North America: an identification manual . American Arachnological Society , Keene, NH . USFWS . 2014 . Michaux’s Sumac (Rhus michauxii) 5-year review . U.S. Fish and Wildlife Service, Southeast Region, Raleigh Ecological Services Field Office , Raleigh, NC . Vamosi J. C. , and S. P. Otto . 2002 . When looks can kill: the evolution of sexually dimorphic floral display and the extinction of dioecious plants . Proc. Biol. Sci . 269 : 1187 – 1194 . Google Scholar Crossref Search ADS PubMed Vamosi J. C. , and S. M. Vamosi . 2004 . The role of diversification in causing the correlates of dioecy . Evolution 58 : 723 – 731 . Google Scholar Crossref Search ADS PubMed Vamosi J. C. , and S. M. Vamosi . 2005 . Present day risk of extinction may exacerbate the lower species richness of dioecious clades . Divers. Distrib . 11 : 25 – 32 . Google Scholar Crossref Search ADS Vázquez D. P. , W. F. Morris , and P. Jordano . 2005 . Interaction frequency as a surrogate for the total effect of animal mutualists on plants . Ecol. Lett . 8 : 1088 – 1094 . Google Scholar Crossref Search ADS Venables W. N. , and B. D. Ripley . 2002 . Modern applied statistics with S , Fourth Edition, Springer , New York . Weakley A. S. 2015 . Flora of the Southern and Mid-Atlantic States, Working Draft of 21 May 2015 , University of North Carolina Herbarium (NCU) Chapel Hill, NC . http://www.herbarium.unc.edu/flora.htm. Williams H. J. , M. R. Strand , G. W. Elzen , S. B. Vinson , and S. J. Merritt . 1986 . Nesting behavior, nest architecture, and use of Dufour’s gland lipids in nest provisioning by Megachile integra and M. mendica mendica (Hymenoptera: Megachilidae) . J. Kans. Entomol. Soc . 59 : 588 – 597 . Ye Q. , E. Bunn , S. L. Krauss , and K. W. Dixon . 2007 . Reproductive success in a reintroduced population of a critically endangered shrub, Symonanthus bancroftii (Solanaceae) . Aust. J. Bot . 55 : 425 – 432 . Google Scholar Crossref Search ADS Young D. A. 1972 . The reproductive biology of Rhus integrifolia and Rhus ovata (Anacardiaceae) . Evolution 26 : 406 – 414 . Google Scholar Crossref Search ADS PubMed Youngsteadt E. , R. E. Irwin , A. Fowler , M. A. Bertone , S. J. Giacomini , M. Kunz , D. Suiter , and C. E. Sorenson . 2018 . Venus flytrap rarely traps its pollinators . Am. Nat . 191 : 539 – 546 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Failure of Pollen Transport Despite High Bee Visitation in an Endangered Dioecious Shrub
JF - Annals of the Entomological Society of America
DO - 10.1093/aesa/say049
DA - 2019-05-07
UR - https://www.deepdyve.com/lp/oxford-university-press/failure-of-pollen-transport-despite-high-bee-visitation-in-an-dDb1TqOoBQ
SP - 169
VL - 112
IS - 3
DP - DeepDyve
ER -